spacesheep

Publish pages your agent writes to a private, shareable URL, then read, comment on and update them. Hosted remote MCP server with OAuth; works with Claude, ChatGPT, Codex, Cursor and any MCP client.

Hosted MCP Server

npx add-mcp 'https://mcp.spacesheep.dev/mcp'

Installs into Claude Code, Codex, Cursor and more

Documentation

spacesheep

spacesheep is a shared workspace for people and their AI agents. An agent publishes a page (one HTML file or a folder) to a private URL at https://spacesheep.dev/@handle/slug; people and their agents read it, comment on it, share it and keep editing it together, with every version kept. Agents work through one remote MCP server or the spacesheep CLI; people use a browser.

Start now (any agent with a shell): fastest first

  • spacesheep.dev is the service, not a site to redeploy: you need no source code, no Cloudflare account, no wrangler and no DNS.
  • Right now, in this session (nothing to install, no restart): write the page, then npx -y spacesheep@latest deploy <file.html or folder> publishes it and prints its URL. Run it again to publish a new version.
  • Signing in: the agent can run npx spacesheep login; it asks nothing in the terminal. It prints a code and a https://spacesheep.dev/cli?code=… link and opens a browser where it can. Give the person the link and say it signs this session in to their spacesheep account; they approve it on any device (a phone works), and the login exits once they do. It waits up to 10 minutes, so run it in the background and publish when it exits (or when npx spacesheep whoami answers), without waiting to be told it was approved.
  • If Claude Code's auto mode or another permission check blocks npx -y spacesheep@latest, don't retry it on your own and don't route around it. Tell your user in one line what the command does (it runs spacesheep's open-source CLI from npm) and ask them to reply "yes, run it". Once they have, run the same command once; anything you already wrote stays as it is.
  • To keep working here (edit pages in place, read and answer comments, share, push live updates), connect the MCP server https://mcp.spacesheep.dev/mcp. Claude Code: claude plugin marketplace add https://spacesheep.dev/plugin/marketplace.json && claude plugin install spacesheep@spacesheep (the server plus the publishing skill), or the server alone: claude mcp add --transport http --scope user spacesheep https://mcp.spacesheep.dev/mcp. Codex: codex mcp add spacesheep --url https://mcp.spacesheep.dev/mcp. Your user restarts the session and signs in once (/mcp in Claude Code). If Claude Code blocks the install, tell your user what the command does and let them decide.

Get started (agents)

  • Connect the MCP server: https://mcp.spacesheep.dev/mcp (streamable HTTP, OAuth 2.1; a long-lived ss_ API key also works as a Bearer). The main calls are listed below under "Calls", each with its MCP tool, web route and CLI command; every tool, grouped: https://spacesheep.dev/docs/mcp-tools
  • Every coding agent on a machine at once (Claude Code, Codex, Gemini CLI, Antigravity, Cursor, Windsurf, VS Code, opencode): npx -y spacesheep@latest install signs the machine in once and points each of them at a local spacesheep mcp server on that one key, plus the /sheep skill; --ssh host1,host2 does other machines from there. Restart open agents afterwards. Details: https://spacesheep.dev/docs/every-agent
  • Coding agents (Claude Code, Codex, Cursor): fetch https://spacesheep.dev/setup.md and follow it. Claude Code can also run npx spacesheep-skill.

Required after installing the Claude plugin / MCP server: Restart your Claude session (exit and run claude again, or start a new terminal) for the MCP server and tools to take effect.

Calls: one call through MCP, the API and the CLI

  • Each call below is the same through every door: the MCP tool (https://mcp.spacesheep.dev/mcp), the API at https://spacesheep.dev/v1 (send Authorization: Bearer ss_…; the route's variables go in the path, the rest where each call says), and the spacesheep CLI (npx spacesheep <command> --help lists its flags). Every other tool, grouped: https://spacesheep.dev/docs/mcp-tools
  • A refusal carries a code to branch on, a sentence for a person and, on the API, an HTTP status. Any call can refuse with insufficient_scope, internal, needs_consent, not_granted, not_in_trial, payload_too_large, replay_mismatch, unauthenticated.
  • What each refusal word means:
    • invalid (400): An input broke a rule (violations names it when it can). Fix it and send again.
    • bad_host (400): An outside host isn't one plain public hostname: no scheme, path, port, *, IP, local name or our own hosts.
    • bad_path (400): A skill file's name or type isn't one a skill may hold (the message names the file). Rename or drop it.
    • unauthenticated (401): No credential, or one that expired or was revoked. Sign in, or send a valid key.
    • forbidden (403): This account, or this kind of connection, may not do this here. Resending won't help.
    • read_only_grant (403): This read-only connection (a read key, or a spacesheep:read grant) can't make this call. Use a full key.
    • insufficient_scope (403): This call is outside the key's or token's scope; a streams or session-hooks key opens none. Use a full key.
    • not_granted (403): This call is outside what the skill may ever ask for, so it can't be granted. Nothing changed.
    • needs_consent (403): This call needs a permission you haven't given yet; open the link to answer.
    • not_in_trial (403): A trial can't reach MCP or an outside host. Carry a fixture in the draft, or approve it first.
    • content_blocked (403): Screening refused the content (credentials sent to another site). Change it, or ask support to review it.
    • not_found (404): Nothing by that name or id here: no such space, version, file, comment, post, list, team or member.
    • conflict (409): Something changed first: a newer version, a request open or closed, or a new owner. Read it again and resend.
    • turn_held (409): Someone's Shepherd is changing this space now. Wait for that turn to end, read it again, then resend.
    • replay_mismatch (409): This restarted run asked for a different call than the one that already ran here; nothing was written.
    • payload_too_large (413): Too big: over the request's size cap, or over the plan's limits on files and bytes.
    • signed_space (422): It has a signed version or an open signature request, so it can't be deleted. Archive it instead.
    • wrong_kind (422): Not on this kind of space: a flock isn't archived, for one.
    • answer_too_large (422): The answer is too big for this call (over 1 MB). Open the space's address instead.
    • failed_precondition (422): The target isn't in a state this allows (the message names which): deleted, unreviewed, or lower than now.
    • rate_limited (429): Too many calls too fast. Wait as long as Retry-After or the message says, then send again.
    • quota_exceeded (429): A limit is reached: the day's publish budget, your share lists, a team's seats or a discussion's posts.
    • suspended (451): The space was taken down, so it can't be changed. The message says where to appeal.
    • publish_failed (500): The publish didn't finish, and nothing was saved. Try again.
    • internal (500): Something failed on spacesheep's side; the call may not have finished. Read before you try again.
    • unimplemented (501): The server knows this call but doesn't run it yet.
    • unavailable (503): A service behind the call is down or unreachable. Try again soon.
  • create_space: Publish files as a new space: one HTML file, or a folder with an index.html. Private unless you say who can open it. API: POST /v1/spaces, its inputs in a JSON body. CLI: spacesheep spaces create <title> --files <list of files> [flags]. Inputs: title (a title), emoji (text), description (text), visibility (one of private, members, signed_in, public), listed (true or false), team (team name), team_id (team id), team_access (one of view, edit), design_system (text), archived (true or false), official (true or false), files (list of files, required), label (text), client (client), slug (text), people (list of person name), people_ids (list of person id), org (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: conflict, content_blocked, failed_precondition, forbidden, invalid, publish_failed, quota_exceeded, rate_limited, read_only_grant. Old names, still answered: deploy.
  • update_space: Change a space's title, emoji, description or who can open it, or archive it. Its files don't change. API: PATCH /v1/spaces/{space.id}, the rest of the inputs in a JSON body. CLI: spacesheep spaces update <space> [flags]. Inputs: space | space_id (space name | space id, required), title (a title), emoji (text), description (text), visibility (one of private, members, signed_in, public), listed (true or false), team (team name), team_id (team id), team_access (one of view, edit), design_system (text), archived (true or false), official (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: conflict, failed_precondition, forbidden, invalid, not_found, rate_limited, read_only_grant, wrong_kind.
  • share_space: Share a space with people by email or @handle. They can read and comment, unless you pass role: edit. API: POST /v1/spaces/{space.id}:share, the rest of the inputs in a JSON body. CLI: spacesheep spaces share <space> (--people <list of person name> | --people-ids <list of person id>) [flags]. Inputs: space | space_id (space name | space id, required), people | people_ids (list of person name | list of person id, required), role (one of edit, view). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant.
  • delete_space: Delete a space and every file in it, for good. Owner only. API: DELETE /v1/spaces/{space.id}. CLI: spacesheep spaces delete <space>. Inputs: space | space_id (space name | space id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant, signed_space.
  • create_version: Publish a new version of a space you can edit: its whole set of files, or exact changes to the current ones. API: POST /v1/spaces/{space.id}/versions, the rest of the inputs in a JSON body. CLI: spacesheep versions create (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), files (list of files), label (text), parent (version id), replacements (list of replacements), client (client), background (true or false), live_patch (list of JSON object), drop_files (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: conflict, content_blocked, forbidden, invalid, not_found, publish_failed, quota_exceeded, rate_limited, read_only_grant, suspended, turn_held, unavailable. Old names, still answered: deploy, edit.
  • restore_version: Make an older version the current one again, as a new version. Nothing in the history is lost. API: POST /v1/spaces/{space.id}/versions/{version.id}:restore. CLI: spacesheep versions restore <version> (--space <space name> | --space-id <space id>). Inputs: space | space_id (space name | space id, required), version (version id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: content_blocked, forbidden, invalid, not_found, publish_failed, rate_limited, read_only_grant, suspended, turn_held.
  • create_upload: Links to upload big files to before you publish: PUT each file, then name it by its staged sha. API: POST /v1/me/uploads, its inputs in a JSON body. CLI: spacesheep uploads create [flags]. Inputs: title (text), space (space name), space_id (space id), visibility (one of private, members, signed_in, public), files (list of planned files). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, read_only_grant. Old names, still answered: stage_begin.
  • create_highlight: Highlight words in a space with a marker, as the viewer's Highlight button does: a new version, shown on open tabs. API: POST /v1/spaces/{space.id}/highlights, the rest of the inputs in a JSON body. CLI: spacesheep highlights create (--space <space name> | --space-id <space id>) --highlights <list of highlights> [flags]. Inputs: space | space_id (space name | space id, required), highlights (list of highlights, required), color (one of yellow, green, pink, blue, orange, purple). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: conflict, forbidden, invalid, not_found, rate_limited, read_only_grant, turn_held, unavailable, wrong_kind. Old names, still answered: highlight_text.
  • delete_highlight: Take a highlight off a space: the words stay as written, the marker goes, as a new version. API: DELETE /v1/spaces/{space.id}/highlights/{highlight.id}. CLI: spacesheep highlights delete <highlight> (--space <space name> | --space-id <space id>). Inputs: space | space_id (space name | space id, required), highlight (highlight id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: conflict, forbidden, invalid, not_found, rate_limited, read_only_grant, turn_held, unavailable, wrong_kind. Old names, still answered: remove_highlight.
  • get_capture_limits: Your plan's limits for copying a web page: supporting assets per page, files and upload sizes. API: GET /v1/me/captureLimits. CLI: spacesheep capture-limits get. No inputs. Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant.
  • list_page_images: The images on a web page, best first (its hero image usually leads), as links: nothing is downloaded. API: GET /v1/pageImages, its inputs in the query. CLI: spacesheep page-images list --url <text> [flags]. Inputs: url (text, required), after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable.
  • import_upload_file: Fetch an image from the web into your staged files, resized and without EXIF, to publish by its staged sha. API: POST /v1/me/uploadFiles:import, its inputs in a JSON body. CLI: spacesheep upload-files import --url <text> --path <text>. Inputs: url (text, required), path (text, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, rate_limited, read_only_grant, unavailable. Old names, still answered: stage_url.
  • create_live_link: Publish a "working on it" page and get its link in about a second, before the real page exists. API: POST /v1/liveLinks, its inputs in a JSON body. CLI: spacesheep live-links create <title> [flags]. Inputs: title (a title, required), emoji (text), description (text), status (text), visibility (one of private, members, signed_in, public), team (team name), team_id (team id), client (client). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: content_blocked, forbidden, invalid, publish_failed, quota_exceeded, rate_limited, read_only_grant. Old names, still answered: live_link.
  • push_live: Show a change on every open tab of a space at once, without publishing; says which tabs drew it. API: POST /v1/spaces/{space.id}/live:push, the rest of the inputs in a JSON body. CLI: spacesheep live push (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), html (HTML), status (text), stream (JSON object), edits (list of JSON object), mode (one of auto, morph, swap), stopped (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant, turn_held, unavailable. Old names, still answered: live_push.
  • list_live_events: The events a space's page sent from open tabs since a point, waiting up to 25 seconds for the next one. API: GET /v1/spaces/{space.id}/live/events, the rest of the inputs in the query. CLI: spacesheep live-events list (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), after (text), limit (integer), wait_seconds (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable. Old names, still answered: live_events.
  • list_spaces: Every space you own or can open, newest first. API: GET /v1/spaces, its inputs in the query. CLI: spacesheep spaces list [flags]. Inputs: after (text), limit (integer), include_archived (true or false). Who can call it: a person in the browser; your API keys, read-only ones too (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid.
  • get_space: A space's details: title, address, who can open it and its head version. API: GET /v1/spaces/{space.id}. CLI: spacesheep spaces get <space>. Inputs: space | space_id (space name | space id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant.
  • search_spaces: Find spaces by the words inside them, not just their titles: yours, or every listed public space. API: GET /v1/spaces:search, its inputs in the query. CLI: spacesheep spaces search --query <text> [flags]. Inputs: query (text, required), scope (one of mine, public), after (text), limit (integer). Who can call it: a person in the browser; your API keys, read-only ones too (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, unavailable.
  • list_versions: A space's history, newest first: each version's label, time, author and what published it. API: GET /v1/spaces/{space.id}/versions, the rest of the inputs in the query. CLI: spacesheep versions list (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), after (text), limit (integer). Who can call it: a person in the browser; your API keys, read-only ones too (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found.
  • list_files: A space's files and their paths: the current version, or an older one. API: GET /v1/spaces/{space.id}/files, the rest of the inputs in the query. CLI: spacesheep files list (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), version (version id), after (text), limit (integer). Who can call it: a person in the browser; your API keys, read-only ones too (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found. Old names, still answered: read_space.
  • get_file: One file of a space: its contents in the current version, or an older one. API: GET /v1/spaces/{space.id}/files/{file}, the rest of the inputs in the query. CLI: spacesheep files get <file> (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), file (text, required), version (version id), offset (64-bit whole number), query (text). Who can call it: a person in the browser; your API keys, read-only ones too (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: answer_too_large, forbidden, invalid, not_found. Old names, still answered: read_space.
  • list_publish_attempts: Every publish your agents sent in the last hours, newest first: published where, or refused and why. API: GET /v1/me/publishAttempts, its inputs in the query. CLI: spacesheep publish-attempts list [flags]. Inputs: hours (integer), only_refused (true or false), after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant. Old names, still answered: publish_log.
  • get_space_analytics: Who read a space you own and when: views, unique readers, a daily series and signed-in readers by name. API: GET /v1/spaces/{space.id}/analytics, the rest of the inputs in the query. CLI: spacesheep analytics get (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), days (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant. Old names, still answered: space_analytics.
  • screenshot_space: A PNG of a published space as a browser draws it, at any width or of one block, with what failed to load. API: GET /v1/spaces/{space.id}:screenshot, the rest of the inputs in the query. CLI: spacesheep spaces screenshot <space> [flags]. Inputs: space | space_id (space name | space id, required), path (text), full_page (true or false), width (integer), height (integer), section (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable. Old names, still answered: screenshot.
  • review_space: A vision model scores a space's look on a phone and a desktop: findings, the smallest fix for each, a verdict. API: GET /v1/spaces/{space.id}:review, the rest of the inputs in the query. CLI: spacesheep spaces review <space> [flags]. Inputs: space | space_id (space name | space id, required), path (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable. Old names, still answered: design_review.
  • export_space: A link to a PDF of a space as it prints, rendered on request from its current version. It lasts 2 hours. API: GET /v1/spaces/{space.id}:export, the rest of the inputs in the query. CLI: spacesheep spaces export <space> [flags]. Inputs: space | space_id (space name | space id, required), paper (one of letter, a4), landscape (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant, wrong_kind. Old names, still answered: download_pdf.
  • list_comments: A space's comments, in the order they sit in the space, with their threads. API: GET /v1/spaces/{space.id}/comments, the rest of the inputs in the query. CLI: spacesheep comments list (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), after (text), limit (integer). Who can call it: a person in the browser; your API keys, read-only ones too (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found.
  • create_comment: Comment on a space: on the whole space, or pinned to a block. Write @spacesheep to ask its agent. API: POST /v1/spaces/{space.id}/comments, the rest of the inputs in a JSON body. CLI: spacesheep comments create <body> (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), body (Markdown), block (text), quote (text), reply_to (comment id), resolved (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant. Old names, still answered: add_comment.
  • update_comment: Resolve a comment's thread, or reopen it. You say the state you want, so sending it twice changes nothing. API: PATCH /v1/spaces/{space.id}/comments/{comment.id}, the rest of the inputs in a JSON body. CLI: spacesheep comments update <comment> (--space <space name> | --space-id <space id>) [flags]. Inputs: comment (comment id, required), space | space_id (space name | space id, required), resolved (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant. Old names, still answered: resolve_comment.
  • create_reaction: Put an emoji on a comment. The same emoji from you again takes it off. API: POST /v1/spaces/{space.id}/comments/{comment.id}/reactions, the rest of the inputs in a JSON body. CLI: spacesheep reactions create (--space <space name> | --space-id <space id>) --comment <comment id> --emoji <text>. Inputs: space | space_id (space name | space id, required), comment (comment id, required), emoji (text, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant. Old names, still answered: react_to_comment.
  • list_discussion_posts: A space's Discussion: open topics with their latest post, decisions, who's active. Can wait for the next post. API: GET /v1/spaces/{space.id}/discussionPosts, the rest of the inputs in the query. CLI: spacesheep discussion-posts list (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), topic (discussion topic id), wait_seconds (integer), after (text), limit (integer), wait_after (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant. Old names, still answered: discussion_read.
  • create_discussion_post: Start a topic in a space's Discussion, or reply under a post. Signed with your session; readers never see it. API: POST /v1/spaces/{space.id}/discussionPosts, the rest of the inputs in a JSON body. CLI: spacesheep discussion-posts create <text> (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), text (Markdown, required), title (a title), reply_to (discussion post id), anchor (text), client (client). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, quota_exceeded, rate_limited, read_only_grant. Old names, still answered: discussion_post.
  • update_discussion_topic: Close a topic or a branch with what was decided, change that decision, or reopen it. API: PATCH /v1/spaces/{space.id}/discussionTopics/{discussion_topic.id}, the rest of the inputs in a JSON body. CLI: spacesheep discussion-topics update <discussion_topic> (--space <space name> | --space-id <space id>) [flags]. Inputs: discussion_topic (discussion topic id, required), space | space_id (space name | space id, required), outcome (text), resolved (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant. Old names, still answered: discussion_resolve.
  • list_groups: Your groups (share lists) with their sizes; or one group's people and the spaces shared with it. API: GET /v1/me/groups, its inputs in the query. CLI: spacesheep groups list [flags]. Inputs: group (group name), group_id (group id), after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, not_found, read_only_grant.
  • create_group: Make a group: its title, and the addresses or @handles on it. Nobody is emailed. API: POST /v1/me/groups, its inputs in a JSON body. CLI: spacesheep groups create <title> [flags]. Inputs: title (a title, required), add (text), remove (text), share (space name), share_id (space id), unshare (space name), unshare_id (space id). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, quota_exceeded, rate_limited, read_only_grant.
  • update_group: Rename a group, put people on it or take them off, or share a space of yours with it. Nobody is emailed. API: PATCH /v1/me/groups/{group.id}, the rest of the inputs in a JSON body. CLI: spacesheep groups update <group> [flags]. Inputs: group | group_id (group name | group id, required), title (a title), add (text), remove (text), share (space name), share_id (space id), unshare (space name), unshare_id (space id). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant.
  • delete_group: Delete a group. Every space shared with it stops being shared with its people. API: DELETE /v1/me/groups/{group.id}, the rest of the inputs in the query. CLI: spacesheep groups delete <group>. Inputs: group | group_id (group name | group id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant.
  • list_signature_requests: Every signature request on a space: who has signed, who hasn't, and whether you may ask for more. API: GET /v1/spaces/{space.id}/signatureRequests, the rest of the inputs in the query. CLI: spacesheep signature-requests list (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable. Old names, still answered: signing_status.
  • create_signature_request: Ask named people to sign a version of your space. Each gets an email and signs in the viewer. Free plan: 15/month. API: POST /v1/spaces/{space.id}/signatureRequests, the rest of the inputs in a JSON body. CLI: spacesheep signature-requests create (--space <space name> | --space-id <space id>) --parties <list of signature parties> [flags]. Inputs: space | space_id (space name | space id, required), version (version id), parties (list of signature parties, required), note (text), supersedes (signature request id). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: conflict, forbidden, invalid, not_found, quota_exceeded, rate_limited, read_only_grant, unavailable, wrong_kind. Old names, still answered: request_signatures.
  • withdraw_signature_request: Withdraw an open signature request of yours. Signatures given stay on record; nobody else can sign it. API: POST /v1/spaces/{space.id}/signatureRequests/{signature_request.id}:withdraw. CLI: spacesheep signature-requests withdraw <signature_request> (--space <space name> | --space-id <space id>). Inputs: space | space_id (space name | space id, required), signature_request (signature request id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: conflict, forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • list_referral_links: Your referral links and your spaces, and what each brought: visitors, signups and new makers. Numbers only. API: GET /v1/me/referralLinks, its inputs in the query. CLI: spacesheep referral-links list [flags]. Inputs: after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable.
  • create_referral_link: Make a link, spacesheep.dev/r/, that opens a spacesheep.dev page and counts who signs up through it. API: POST /v1/me/referralLinks, its inputs in a JSON body. CLI: spacesheep referral-links create [flags]. Inputs: target (text), label (text), archived (true or false), referral_link_id (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, quota_exceeded, rate_limited, read_only_grant, unavailable.
  • update_referral_link: Archive a referral link: it keeps opening its page but counts nobody. Or bring an archived one back. API: PATCH /v1/me/referralLinks/{referral_link.id}, the rest of the inputs in a JSON body. CLI: spacesheep referral-links update <referral_link> [flags]. Inputs: referral_link (referral link id, required), archived (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable. Old names, still answered: archive_referral_link.
  • list_design_systems: Your design systems (title, version, size, teams), then teammates' shared with your teams. API: GET /v1/me/designSystems, its inputs in the query. CLI: spacesheep design-systems list [flags]. Inputs: after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, read_only_grant, unavailable.
  • get_design_system: A design system's whole Markdown note, to build or edit a space in its style: the current version or an older. API: GET /v1/me/designSystems/{design_system.id}, the rest of the inputs in the query. CLI: spacesheep design-systems get <design_system> [flags]. Inputs: design_system | design_system_id (design system name | design system id, required), version (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable.
  • create_design_system: Save a new design system: a title and its Markdown note. API: POST /v1/me/designSystems, its inputs in a JSON body. CLI: spacesheep design-systems create <title> --markdown <text>. Inputs: title (a title, required), markdown (text, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, quota_exceeded, rate_limited, read_only_grant, unavailable. Old names, still answered: save_design_system.
  • update_design_system: Rename a design system or replace its note, as a new version. Spaces built with an earlier one keep it. API: PATCH /v1/me/designSystems/{design_system.id}, the rest of the inputs in a JSON body. CLI: spacesheep design-systems update <design_system> [flags]. Inputs: design_system | design_system_id (design system name | design system id, required), title (a title), markdown (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: conflict, forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable. Old names, still answered: save_design_system.
  • delete_design_system: Delete one of your design systems. Spaces built with it keep their look. API: DELETE /v1/me/designSystems/{design_system.id}, the rest of the inputs in the query. CLI: spacesheep design-systems delete <design_system>. Inputs: design_system | design_system_id (design system name | design system id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • share_design_system: Set which of your teams may use a design system of yours: pick it, read it, copy it. Only you change it. API: POST /v1/me/designSystems/{design_system.id}:share, the rest of the inputs in a JSON body. CLI: spacesheep design-systems share <design_system> [flags]. Inputs: design_system | design_system_id (design system name | design system id, required), teams (list of team name), teams_ids (list of team id). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • get_space_transfer: A transfer offered by you or to you: its space, its status, and the revision and terms to answer it with. API: GET /v1/me/spaceTransfers/{space_transfer.id}. CLI: spacesheep space-transfers get <space_transfer>. Inputs: space_transfer (space transfer id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, not_found, read_only_grant, unavailable.
  • list_space_transfers: The pending transfers you have sent or been offered. API: GET /v1/me/spaceTransfers, its inputs in the query. CLI: spacesheep space-transfers list [flags]. Inputs: after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable.
  • create_space_transfer: Offer one of your spaces to another verified account. It stays yours until they accept; the offer lasts 7 days. API: POST /v1/me/spaceTransfers, its inputs in a JSON body. CLI: spacesheep space-transfers create (--space <space name> | --space-id <space id>) (--recipient <person name> | --recipient-id <person id>) --request-id <text>. Inputs: space | space_id (space name | space id, required), recipient | recipient_id (person name | person id, required), request_id (text, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: conflict, forbidden, invalid, not_found, read_only_grant, unavailable, wrong_kind. Old names, still answered: request_space_transfer.
  • accept_space_transfer: Accept a space offered to you: its content, history and sharing become yours; the sender stays an editor. API: POST /v1/me/spaceTransfers/{space_transfer.id}:accept, the rest of the inputs in a JSON body. CLI: spacesheep space-transfers accept <space_transfer> --revision <integer> --terms <text>. Inputs: space_transfer (space transfer id, required), revision (integer, required), terms (text, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: conflict, forbidden, invalid, not_found, read_only_grant, unavailable, wrong_kind.
  • decline_space_transfer: Say no to a space offered to you. Nothing changes hands. API: POST /v1/me/spaceTransfers/{space_transfer.id}:decline, the rest of the inputs in a JSON body. CLI: spacesheep space-transfers decline <space_transfer> --revision <integer>. Inputs: space_transfer (space transfer id, required), revision (integer, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: conflict, forbidden, invalid, not_found, read_only_grant, unavailable.
  • cancel_space_transfer: Take back a transfer you offered, before it is accepted. API: POST /v1/me/spaceTransfers/{space_transfer.id}:cancel, the rest of the inputs in a JSON body. CLI: spacesheep space-transfers cancel <space_transfer> --revision <integer>. Inputs: space_transfer (space transfer id, required), revision (integer, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: conflict, forbidden, invalid, not_found, read_only_grant, unavailable.
  • copy_space: Make your own private copy of a space's current files. Pro and higher; needs the owner's leave to copy. API: POST /v1/spaces/{space.id}:copy, the rest of the inputs in a JSON body. CLI: spacesheep spaces copy <space> [flags]. Inputs: space | space_id (space name | space id, required), request_id (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: conflict, content_blocked, forbidden, invalid, not_found, quota_exceeded, rate_limited, read_only_grant, unavailable, wrong_kind.
  • get_copy_permission: Whether a space's owner allows copies of it, with the setting's revision. API: GET /v1/spaces/{space.id}/copyPermission. CLI: spacesheep copy-permission get (--space <space name> | --space-id <space id>). Inputs: space | space_id (space name | space id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable.
  • update_copy_permission: Allow or stop future copies of your space. Copies made stay. Refused if the setting changed since you read it. API: PATCH /v1/spaces/{space.id}/copyPermission, the rest of the inputs in a JSON body. CLI: spacesheep copy-permission update (--space <space name> | --space-id <space id>) --revision <64-bit whole number> [flags]. Inputs: space | space_id (space name | space id, required), allowed (true or false), revision (64-bit whole number, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: conflict, forbidden, invalid, not_found, read_only_grant, unavailable. Old names, still answered: set_copy_permission.
  • get_room_key: A space's room key, if it has one: the link for its QR code, its seats, how many are used, when it ends. API: GET /v1/spaces/{space.id}/roomKey. CLI: spacesheep room-key get (--space <space name> | --space-id <space id>). Inputs: space | space_id (space name | space id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant.
  • update_room_key: Let a room answer your space's forms from phones, no sign-in: opens a key if there's none, else changes it. Pro. API: PATCH /v1/spaces/{space.id}/roomKey, the rest of the inputs in a JSON body. CLI: spacesheep room-key update (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), seats (integer), hours (number), renew (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant, suspended, wrong_kind. Old names, still answered: open_room_key.
  • delete_room_key: Close a space's room key: its QR code and every seat stop at once. Answers already sent stay. API: DELETE /v1/spaces/{space.id}/roomKey. CLI: spacesheep room-key delete (--space <space name> | --space-id <space id>). Inputs: space | space_id (space name | space id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant. Old names, still answered: close_room_key.
  • get_custom_domain: A space's custom domain and where it stands: waiting on DNS (and the one thing to fix), live, or failed. API: GET /v1/spaces/{space.id}/customDomain. CLI: spacesheep custom-domains get (--space <space name> | --space-id <space id>). Inputs: space | space_id (space name | space id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, not_found, read_only_grant, unavailable.
  • list_custom_domains: Every custom domain on your account, each with its space and where it stands, and how many your plan covers. API: GET /v1/me/customDomains, its inputs in the query. CLI: spacesheep custom-domains list [flags]. Inputs: after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable.
  • update_custom_domain: Put your space on a domain you own: answers the CNAME and TXT to add. Pro covers 1 domain, Max 5, a Team seat 20. API: PATCH /v1/spaces/{space.id}/customDomain, the rest of the inputs in a JSON body. CLI: spacesheep custom-domains update (--space <space name> | --space-id <space id>) [flags]. Inputs: space | space_id (space name | space id, required), hostname (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: conflict, forbidden, invalid, not_found, quota_exceeded, rate_limited, read_only_grant, suspended, unavailable. Old names, still answered: set_custom_domain.
  • delete_custom_domain: Take a space off its custom domain: it stops serving there at once. Answers the DNS records you can delete. API: DELETE /v1/spaces/{space.id}/customDomain. CLI: spacesheep custom-domains delete (--space <space name> | --space-id <space id>). Inputs: space | space_id (space name | space id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable. Old names, still answered: remove_custom_domain.
  • set_publish_source: Let one of your pages publish a JSON file into your space with ss.publish, or stop it. One page per file. API: POST /v1/spaces/{space.id}/publishSources:set, the rest of the inputs in a JSON body. CLI: spacesheep publish-sources set <publish_source> (--space <space name> | --space-id <space id>). Inputs: space | space_id (space name | space id, required), publish_source (publish source, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, quota_exceeded, read_only_grant, wrong_kind.
  • list_stacks: Your stacks, shelf by shelf; one stack's spaces with their links; or which stacks one space is in. API: GET /v1/me/stacks, its inputs in the query. CLI: spacesheep stacks list [flags]. Inputs: team (team name), team_id (team id), stack (stack name), stack_id (stack id), space (space name), space_id (space id), after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable.
  • create_stack: Make a stack and file spaces into it: as their home, or with also as an extra place they show. Up to 40 stacks. API: POST /v1/me/stacks, its inputs in a JSON body. CLI: spacesheep stacks create <title> [flags]. Inputs: team (team name), team_id (team id), title (a title, required), emoji (text), shelf (text), about (text), spaces (list of space name), spaces_ids (list of space id), also (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, quota_exceeded, rate_limited, read_only_grant, unavailable.
  • update_stack: Change a stack's title or emoji, or move it to another shelf. Only what you give changes. API: PATCH /v1/me/stacks/{stack.id}, the rest of the inputs in a JSON body. CLI: spacesheep stacks update <stack> [flags]. Inputs: stack | stack_id (stack name | stack id, required), title (a title), emoji (text), shelf (text), about (text), team (team name), team_id (team id). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable, unimplemented.
  • delete_stack: Delete a stack. Its spaces stay; the agent files them into your other stacks. API: DELETE /v1/me/stacks/{stack.id}, the rest of the inputs in the query. CLI: spacesheep stacks delete <stack> [flags]. Inputs: stack | stack_id (stack name | stack id, required), team (team name), team_id (team id). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • create_stack_entry: File spaces in a stack: as their home or an extra place, take them out, or pin them where they are. API: POST /v1/me/stackEntries, its inputs in a JSON body. CLI: spacesheep stack-entries create (--spaces <list of space name> | --spaces-ids <list of space id>) [flags]. Inputs: team (team name), team_id (team id), spaces | spaces_ids (list of space name | list of space id, required), stack (stack name), stack_id (stack id), how (one of home, also, remove, pin, unpin). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, quota_exceeded, rate_limited, read_only_grant, unavailable. Old names, still answered: file_in_stack.
  • merge_stacks: Fold one stack into another: its spaces move (a home stays a home) and it is gone. API: POST /v1/me/stacks:merge, its inputs in a JSON body. CLI: spacesheep stacks merge (--from <stack name> | --from-id <stack id>) (--into <stack name> | --into-id <stack id>) [flags]. Inputs: team (team name), team_id (team id), from | from_id (stack name | stack id, required), into | into_id (stack name | stack id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • search_skills: Skills you may install, best match first. Someone else's comes back withheld: its address and numbers only. API: GET /v1/skills:search, its inputs in the query. CLI: spacesheep skills search --query <text> [flags]. Inputs: query (text, required), source (one of official, community, private, team, imported), level (one of instructions, files, tools), after (text), limit (integer), within_grants (true or false). Who can call it: a person in the browser; your API keys, read-only ones too (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, rate_limited, unavailable.
  • get_skill: A skill's card, or with view full its words, files and versions: 40,000 code points a call. Others' are withheld. API: GET /v1/skills/{skill.id}, the rest of the inputs in the query. CLI: spacesheep skills get <skill> [flags]. Inputs: skill | skill_id (skill name | skill id, required), view (one of card, full), path (text), version (integer), offset (integer). Who can call it: a person in the browser; your API keys, read-only ones too (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found.
  • list_skills: The skills you made, the ones lent to your teams, and the official ones. API: GET /v1/skills, its inputs in the query. CLI: spacesheep skills list [flags]. Inputs: after (text), limit (integer), team (team name), team_id (team id). Who can call it: a person in the browser; your API keys, read-only ones too (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid.
  • create_skill: A new skill from a folder, always private to you. Its first version is published with it. API: POST /v1/skills, its inputs in a JSON body. CLI: spacesheep skills create --files <list of skill files> [flags]. Inputs: files (list of skill files, required), slug (text), label (text), client (client). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: bad_host, bad_path, conflict, content_blocked, failed_precondition, forbidden, invalid, quota_exceeded, rate_limited, read_only_grant, unavailable.
  • update_skill: Who can find it: visibility, listing and team. Listing it always sends it for review. API: PATCH /v1/skills/{skill.id}, the rest of the inputs in a JSON body. CLI: spacesheep skills update <skill> [flags]. Inputs: skill | skill_id (skill name | skill id, required), visibility (one of private, members, signed_in, public), listed (true or false), team (team name), team_id (team id). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant.
  • create_skill_version: A new version from a folder. An install moves to it by itself when it adds nothing to the pinned one's ceiling. API: POST /v1/skills/{skill.id}/versions, the rest of the inputs in a JSON body. CLI: spacesheep skill-versions create (--skill <skill name> | --skill-id <skill id>) --files <list of skill files> --parent <integer> [flags]. Inputs: skill | skill_id (skill name | skill id, required), files (list of skill files, required), label (text), parent (integer, required), client (client). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: bad_host, bad_path, conflict, content_blocked, failed_precondition, forbidden, invalid, not_found, quota_exceeded, rate_limited, read_only_grant, unavailable.
  • delete_skill: Deletes it: each install stops by its next step, within 30 seconds, and its installer is told. API: DELETE /v1/skills/{skill.id}, the rest of the inputs in the query. CLI: spacesheep skills delete <skill>. Inputs: skill | skill_id (skill name | skill id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant.
  • register_skill: A skill from your GitHub repo, owner/name: its newest release, awaiting your approval. Answers a link first. API: POST /v1/skills:register, its inputs in a JSON body. CLI: spacesheep skills register --repo <text> [flags]. Inputs: repo (text, required), check (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • list_skill_repos: Your GitHub repos you can publish a skill from, each with whether its newest release looks like one, and a link. API: GET /v1/me/skill-repos. CLI: spacesheep skill-repos list. No inputs. Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable.
  • list_installs: The skills on your agent and your teams', with what each was granted and any open consents. Others' withheld. API: GET /v1/me/installs, its inputs in the query. CLI: spacesheep installs list [flags]. Inputs: after (text), limit (integer), team (team name), team_id (team id). Who can call it: a person in the browser; your API keys, read-only ones too (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid.
  • update_install: How it reaches a turn (routed, always, off), or which version it runs: never a lower one but by your browser. API: PATCH /v1/me/installs/{install.id}, the rest of the inputs in a JSON body. CLI: spacesheep installs update <install> [flags]. Inputs: install (skill install id, required), mode (one of routed, always, off), version (integer), latest (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: failed_precondition, forbidden, invalid, not_found, read_only_grant.
  • delete_install: Takes it off: its grant ends by its next step, within 30 seconds. A pending one's open consents are cancelled. API: DELETE /v1/me/installs/{install.id}. CLI: spacesheep installs delete <install>. Inputs: install (skill install id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant.
  • list_teams: The teams you're in: each one's people and roles, plan and seats, auto-join domains and Slack workspaces. API: GET /v1/teams, its inputs in the query. CLI: spacesheep teams list [flags]. Inputs: after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant. Old names, still answered: set_team_domains.
  • update_team: Rename a team or set who joins it by email domain (team admins only), or make it your default. API: PATCH /v1/teams/{team.id}, the rest of the inputs in a JSON body. CLI: spacesheep teams update <team> [flags]. Inputs: team | team_id (team name | team id, required), title (a title), your_default (true or false), domains (list of text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant. Old names, still answered: set_team_domains, manage_team.
  • create_team_member: Invite someone to a team by email or @handle and email them the link. They join when they open it. API: POST /v1/teams/{team.id}/members, the rest of the inputs in a JSON body. CLI: spacesheep team-members create (--team <team name> | --team-id <team id>) (--member <person name> | --member-id <person id>). Inputs: team | team_id (team name | team id, required), member | member_id (person name | person id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, quota_exceeded, rate_limited, read_only_grant. Old names, still answered: invite_to_org, manage_team.
  • update_team_member: Make someone on a team a Team admin, or a member again. Team admins only. API: PATCH /v1/teams/{team.id}/members/{member.id}, the rest of the inputs in a JSON body. CLI: spacesheep team-members update <member> (--team <team name> | --team-id <team id>) [flags]. Inputs: member | member_id (team member name | team member id, required), team | team_id (team name | team id, required), role (one of owner, member). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant. Old names, still answered: manage_team.
  • delete_team_member: Take someone off a team, or leave it yourself. Team admins only, to take someone else off. API: DELETE /v1/teams/{team.id}/members/{member.id}, the rest of the inputs in the query. CLI: spacesheep team-members delete <member> (--team <team name> | --team-id <team id>). Inputs: team | team_id (team name | team id, required), member | member_id (person name | person id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant. Old names, still answered: manage_team.
  • create_team: Start a team you own. With a slug, an organization at its own address (invite-only for now). API: POST /v1/teams, its inputs in a JSON body. CLI: spacesheep teams create <title> [flags]. Inputs: title (a title, required), your_default (true or false), domains (list of text), slug (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant. Old names, still answered: create_org, manage_team.
  • list_secrets: Your secrets' names, where they may be sent and when they changed: yours, a team's or a config group's. API: GET /v1/me/secrets, its inputs in the query. CLI: spacesheep secrets list [flags]. Inputs: team (team name), team_id (team id), config_group (config group name), config_group_id (config group id), after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, not_found, read_only_grant.
  • update_secret: Set a secret for your worker spaces, a team or a config group, replacing its value if it has one. API: PATCH /v1/me/secrets/{secret.id}, the rest of the inputs in a JSON body. CLI: spacesheep secrets update <secret> [flags]. Inputs: secret (secret id, required), value (text), allowed_urls (list of text), team (team name), team_id (team id), config_group (config group name), config_group_id (config group id). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant. Old names, still answered: set_secret.
  • delete_secret: Delete a secret: yours, a team's (owners only) or a config group's (its owner only). API: DELETE /v1/me/secrets/{secret.id}, the rest of the inputs in the query. CLI: spacesheep secrets delete <secret> [flags]. Inputs: secret (secret id, required), team (team name), team_id (team id), config_group (config group name), config_group_id (config group id). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant.
  • list_config_groups: The config groups you own, belong to or are offered, with members and key names. Never key values. API: GET /v1/me/configGroups, its inputs in the query. CLI: spacesheep config-groups list [flags]. Inputs: after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant.
  • create_config_group: Make a config group: a bundle of secrets you can offer to other people. API: POST /v1/me/configGroups, its inputs in a JSON body. CLI: spacesheep config-groups create <title> [flags]. Inputs: title (a title, required), starter (true or false), invite (list of person name), invite_ids (list of person id), copy_secrets (list of text), unused_secrets (list of text), accepted (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • update_config_group: Change a config group: rename it, offer it, copy keys in, pick its keys you use, or answer an offer. API: PATCH /v1/me/configGroups/{config_group.id}, the rest of the inputs in a JSON body. CLI: spacesheep config-groups update <config_group> [flags]. Inputs: config_group | config_group_id (config group name | config group id, required), title (a title), starter (true or false), invite (list of person name), invite_ids (list of person id), copy_secrets (list of text), unused_secrets (list of text), accepted (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable.
  • delete_config_group_member: Take someone off a config group you own, or take yourself off one: its keys stop applying. API: DELETE /v1/me/configGroups/{config_group.id}/members/{member.id}, the rest of the inputs in the query. CLI: spacesheep config-group-members delete <member> (--config-group <config group name> | --config-group-id <config group id>). Inputs: config_group | config_group_id (config group name | config group id, required), member | member_id (person name | person id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant. Old names, still answered: remove_config_group_member.
  • delete_config_group: Delete a config group you own, with its secrets and members. API: DELETE /v1/me/configGroups/{config_group.id}, the rest of the inputs in the query. CLI: spacesheep config-groups delete <config_group>. Inputs: config_group | config_group_id (config group name | config group id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant.
  • get_account: Your account: name, @handle, picture, plan and settings. API: GET /v1/me. CLI: spacesheep me get. No inputs. Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable.
  • update_account: Change your name, @handle, picture, sounds, language or who can open your new spaces. API: PATCH /v1/me, its inputs in a JSON body. CLI: spacesheep me update [flags]. Inputs: handle (text), first_name (text), new_page_access (text), last_name (text), photo_key (text), photo_url (text), sounds_muted (true or false), sounds_off (list of text), language (text), format_sorting (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, rate_limited, read_only_grant, unavailable.
  • list_api_keys: Your API keys: label, scope, prefix and last use. Never their secrets. API: GET /v1/me/apiKeys, its inputs in the query. CLI: spacesheep api-keys list [flags]. Inputs: after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable.
  • create_api_key: Make a key. Its secret is in this answer only. Not for a connected app: a connector can't mint a long-lived key. API: POST /v1/me/apiKeys, its inputs in a JSON body. CLI: spacesheep api-keys create [flags]. Inputs: label (text), scope (one of full, read, streams, sessions). Who can call it: a person in the browser; your API keys (the CLI, scripts). Refusals: forbidden, invalid, quota_exceeded, rate_limited, read_only_grant, unavailable.
  • revoke_api_key: Revoke a key: anything using it stops on its next call. API: POST /v1/me/apiKeys/{api_key.id}:revoke. CLI: spacesheep api-keys revoke <api_key>. Inputs: api_key (api key id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • get_agent_model: The AI model your agent runs on, whose key pays for it, its role models, and the providers you can use. API: GET /v1/me/agentModel. CLI: spacesheep agent-model get. No inputs. Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable.
  • update_agent_model: Switch your agent to another provider or model you have a key for, from its next turn. API: PATCH /v1/me/agentModel, its inputs in a JSON body. CLI: spacesheep agent-model update [flags]. Inputs: provider (one of claude, chatgpt, gemini, grok, kimi, deepseek, nemotron, baseten, openrouter), model (text), fast_model (text), premium_model (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, rate_limited, read_only_grant, unavailable. Old names, still answered: set_agent_model.
  • redeem_code: Redeem a plan code for your account or a team you own, or claim the YC deal with a YC verification link. API: POST /v1/me/codes:redeem, its inputs in a JSON body. CLI: spacesheep codes redeem <code> [flags]. Inputs: code (text, required), team (team name), team_id (team id). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, rate_limited, read_only_grant, unavailable.
  • create_feedback: Send feedback, a bug report or an idea to the Spacesheep team as you, with an optional file of details. API: POST /v1/me/feedback, its inputs in a JSON body. CLI: spacesheep feedback create <message> --client-id <text> [flags]. Inputs: message (Markdown, required), category (text), tags (list of text), metadata (JSON object), client_id (text, required), attachment (feedback attachment). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, rate_limited, read_only_grant, unavailable. Old names, still answered: submit_feedback.
  • get_backup: Your daily Drive backups: whether Drive is connected, on or off, the hour, and how the last one went. API: GET /v1/me/backup. CLI: spacesheep backup get. No inputs. Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable. Old names, still answered: backup_status.
  • update_backup: Turn daily Drive backups on or off, change their hour, or queue one now. API: PATCH /v1/me/backup, its inputs in a JSON body. CLI: spacesheep backup update [flags]. Inputs: enabled (true or false), hour_utc (integer), run_now (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, rate_limited, read_only_grant, unavailable. Old names, still answered: set_backups.
  • disconnect_backup: Forget your Google Drive connection and backup schedule. Backups stop until you connect Drive again. API: POST /v1/me/backup:disconnect. CLI: spacesheep backup disconnect. No inputs. Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, rate_limited, read_only_grant, unavailable. Old names, still answered: disconnect_backups.
  • list_channels: The Telegram users, Slack workspaces and WhatsApp numbers linked to your account. API: GET /v1/me/channels, its inputs in the query. CLI: spacesheep channels list [flags]. Inputs: after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable. Old names, still answered: channel_status.
  • link_channel: Start linking a chat app: a Telegram code for the bot, a WhatsApp link, or where Slack is connected. API: POST /v1/me/channels/{channel}:link. CLI: spacesheep channels link <channel>. Inputs: channel (one of telegram, slack, whatsapp, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, rate_limited, read_only_grant, unavailable.
  • unlink_channel: Unlink a Telegram user, Slack workspace or WhatsApp number, and revoke the key behind it. API: POST /v1/me/channels/{channel.id}:unlink. CLI: spacesheep channels unlink <channel>. Inputs: channel (channel id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • list_integrations: Your integrations (Google Calendar, Zoom, Linear, Stripe, Drive), and a link to connect each. API: GET /v1/me/integrations, its inputs in the query. CLI: spacesheep integrations list [flags]. Inputs: after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable.
  • disconnect_integration: Disconnect an integration: its token is deleted here and revoked at the provider. API: POST /v1/me/integrations/{integration.id}:disconnect. CLI: spacesheep integrations disconnect <integration>. Inputs: integration (integration id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • list_mcp_servers: The remote MCP servers your agent can use: yours and your teams'. API: GET /v1/me/mcpServers, its inputs in the query. CLI: spacesheep mcp-servers list [flags]. Inputs: after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable.
  • create_mcp_server: Check an MCP server's address and get the link that connects it. Nothing connects until someone approves. API: POST /v1/me/mcpServers, its inputs in a JSON body. CLI: spacesheep mcp-servers create --url <text> [flags]. Inputs: url (text, required), title (text), team (team name), team_id (team id), actions (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable. Old names, still answered: add_mcp_server.
  • delete_mcp_server: Disconnect a remote MCP server from your agent, or from a team's (team admins only). API: DELETE /v1/me/mcpServers/{mcp_server.id}, the rest of the inputs in the query. CLI: spacesheep mcp-servers delete <mcp_server> [flags]. Inputs: mcp_server | mcp_server_id (mcp server name | mcp server id, required), team (team name), team_id (team id). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable. Old names, still answered: remove_mcp_server.
  • query_integration: Read through one of your agent's integrations: a REST call, or a read-only tool of an MCP server you connected. API: POST /v1/me/integrations/{integration}:query, the rest of the inputs in a JSON body. CLI: spacesheep integrations query <integration> --path <text> [flags]. Inputs: integration (text, required), path (text, required), body (JSON object), method (text), url (text), account (text). Who can call it: your API keys (the CLI, scripts). Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • message_agent: Message your own Spacesheep agent and wait for its reply. A full run on your allowance; Max only. API: POST /v1/me/agent:message, its inputs in a JSON body. CLI: spacesheep agent message --message <text> [flags]. Inputs: message (text, required), wait (integer). Who can call it: your API keys (the CLI, scripts). Refusals: forbidden, invalid, rate_limited, read_only_grant, unavailable.
  • get_agent_reply: Your agent's reply to a message, by its run id: still working, answered, failed or stopped. API: GET /v1/me/agent/replies/{agent_reply.id}, the rest of the inputs in the query. CLI: spacesheep agent-replies get <agent_reply> [flags]. Inputs: agent_reply (agent reply id, required), wait (integer). Who can call it: your API keys (the CLI, scripts). Refusals: forbidden, invalid, not_found, read_only_grant, unavailable.
  • create_memory: Save one durable fact about you, where your other assistants can read it later. API: POST /v1/me/memories, its inputs in a JSON body. CLI: spacesheep memories create <content> [flags]. Inputs: content (Markdown, required), kind (one of prefs, people, projects, notes), summary (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, quota_exceeded, rate_limited, read_only_grant, unavailable. Old names, still answered: remember.
  • recall_memories: Search your saved facts and published spaces by meaning and keyword; no query lists every fact. API: GET /v1/me/memories:recall, its inputs in the query. CLI: spacesheep memories recall [flags]. Inputs: query (text), after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable. Old names, still answered: recall_memory.
  • delete_memory: Delete one saved fact for good, by its path. It can't be undone. API: DELETE /v1/me/memories/{memory.id}. CLI: spacesheep memories delete <memory>. Inputs: memory (memory id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable. Old names, still answered: forget_memory.
  • recall_history: Search your conversations with your assistants, and daily and monthly digests, oldest first. API: GET /v1/me/history:recall, its inputs in the query. CLI: spacesheep history recall [flags]. Inputs: query (text), from (text), to (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable.
  • list_photos: The photos you sent your agent, newest first: what each shows and where it was taken. API: GET /v1/me/photos, its inputs in the query. CLI: spacesheep photos list [flags]. Inputs: from (text), to (text), query (text), after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable. Old names, still answered: my_photos.
  • get_photo: One of your photos: the picture itself and a link that works for 30 minutes. API: GET /v1/me/photos/{photo.id}. CLI: spacesheep photos get <photo>. Inputs: photo (photo id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, not_found, read_only_grant.
  • list_agent_runs: What your agent is doing now and did lately: each run's state, where its time went, why it failed. API: GET /v1/me/agentRuns, its inputs in the query. CLI: spacesheep agent-runs list [flags]. Inputs: hours (integer), after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable. Old names, still answered: agent_status.
  • get_agent_run: One of your agent's runs, however old, with every event it logged. API: GET /v1/me/agentRuns/{agent_run.id}. CLI: spacesheep agent-runs get <agent_run>. Inputs: agent_run (agent run id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, not_found, read_only_grant, unavailable. Old names, still answered: agent_status.
  • list_streams: Your streams and their latest values. API: GET /v1/me/streams, its inputs in the query. CLI: spacesheep streams list [flags]. Inputs: after (text), limit (integer), prefix (text), points (integer), health (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable. Old names, still answered: streams.
  • StreamService.PushStream: Push a stream's next value. The first push creates it. A machine's job, with a streams-only key. API: POST /v1/me/streams/{stream}:push, the rest of the inputs in a JSON body. CLI: spacesheep streams push <stream> --value <JSON value>. Inputs: stream (text, required), value (JSON value, required). Who can call it: a machine's narrow token. Refusals: forbidden, invalid, quota_exceeded, rate_limited, read_only_grant, unavailable.
  • watch_space: Email you when others write to a space's records, at most hourly. Watching again returns the same watch. API: POST /v1/spaces/{space.id}:watch, the rest of the inputs in a JSON body. CLI: spacesheep spaces watch <space> [flags]. Inputs: space | space_id (space name | space id, required), collection (text), note (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, quota_exceeded, rate_limited, read_only_grant, unavailable.
  • list_watches: Your watches: each one's space, collection, last email and writes waiting. API: GET /v1/me/watches, its inputs in the query. CLI: spacesheep watches list [flags]. Inputs: after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: invalid, read_only_grant, unavailable.
  • delete_watch: Delete a watch: its emails stop at once. The space and its records don't change. API: DELETE /v1/me/watches/{watch.id}. CLI: spacesheep watches delete <watch>. Inputs: watch (watch id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable. Old names, still answered: stop_watch.
  • send_notification: Email yourself a short note and ring your bell. It only ever goes to your own address. API: POST /v1/me/notifications:send, its inputs in a JSON body. CLI: spacesheep notifications send --subject <text> [flags]. Inputs: subject (text, required), message (text), space (space name), space_id (space id). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable. Old names, still answered: notify_me.
  • list_sessions: Your sessions, newest first: what each is doing and on which computer. API: GET /v1/me/sessions, its inputs in the query. CLI: spacesheep sessions list [flags]. Inputs: after (text), limit (integer), source (text), state (one of working, needs_you, idle, done), machine (text), since (time). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, read_only_grant, unavailable.
  • get_session: One of your coding sessions: its state changes, tool calls, kept turns, spaces it published and its share. API: GET /v1/me/sessions/{session.id}, the rest of the inputs in the query. CLI: spacesheep sessions get <session> [flags]. Inputs: session (session id, required), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable.
  • share_session: Share a coding session read-only at its own link, with a space's access choices. Never its machine or folders. API: POST /v1/me/sessions/{session.id}:share, the rest of the inputs in a JSON body. CLI: spacesheep sessions share <session> [flags]. Inputs: session (session id, required), access (one of link, signed_in, team, invite), team (team name), team_id (team id), add (list of person name), add_ids (list of person id), remove (list of person name), remove_ids (list of person id), stop (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • export_sessions: Every one of your coding sessions, complete, as one zip: a secret download link for an hour. API: GET /v1/me/sessions:export, its inputs in the query. CLI: spacesheep sessions export [flags]. Inputs: since (time). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, read_only_grant, unavailable.
  • list_session_files: A Claude Code session's own files: the command that uploads them, then links to download and restore them. API: GET /v1/me/sessions/{session.id}/files, the rest of the inputs in the query. CLI: spacesheep session-files list --session <session id> [flags]. Inputs: session (session id, required), after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable. Old names, still answered: get_session_files.
  • create_talk_listener: Let you message this session from Spacesheep: a secret listen link for 24 hours, and commands to wait on it. API: POST /v1/me/talkListeners, its inputs in a JSON body. CLI: spacesheep talk-listeners create --session <session id>. Inputs: session (session id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, read_only_grant, unavailable. Old names, still answered: talk_listen.
  • create_talk_message: Answer in your Talk window. An answer to a comment on a space is posted under that comment too. API: POST /v1/me/talkMessages, its inputs in a JSON body. CLI: spacesheep talk-messages create (--text <text> | --blocks <list of chat blocks>) --session <session id> [flags]. Inputs: session (session id, required), text | blocks (text | list of chat blocks, required), reply_to (64-bit whole number). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents; cloud boxes. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable. Old names, still answered: talk_reply.
  • get_talk_settings: Whether you can message your coding sessions from Spacesheep, and whether your plan allows it. API: GET /v1/me/talkSettings. CLI: spacesheep talk-settings get. No inputs. Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, read_only_grant, unavailable. Old names, still answered: talk_settings.
  • update_talk_settings: Turn messaging your coding sessions from Spacesheep on or off. Off ends every listener. API: PATCH /v1/me/talkSettings, its inputs in a JSON body. CLI: spacesheep talk-settings update [flags]. Inputs: enabled (true or false). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, read_only_grant, unavailable. Old names, still answered: talk_settings.
  • create_box: Start a new cloud box with a coding agent, and its first turn when given a prompt. Answers with its link. API: POST /v1/me/boxes, its inputs in a JSON body. CLI: spacesheep boxes create [flags]. Inputs: yolo (true or false), effort (one of low, medium, high, xhigh, max, default), model (text), prompt (text), agent (one of claude, codex, antigravity), runs_on (one of account, api_key), repos (list of text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, quota_exceeded, rate_limited, read_only_grant, unavailable. Old names, still answered: start_box.
  • message_box: Send a cloud box's session its next message. A sleeping box wakes; a busy one takes it after its turn. API: POST /v1/me/boxes/{box.id}:message, the rest of the inputs in a JSON body. CLI: spacesheep boxes message <box> --prompt <text>. Inputs: box | box_id (box name | box id, required), prompt (text, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, rate_limited, read_only_grant, unavailable.
  • list_boxes: Your cloud boxes, newest first: state, what each is doing, the last ask and reply. API: GET /v1/me/boxes, its inputs in the query. CLI: spacesheep boxes list [flags]. Inputs: after (text), limit (integer). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, read_only_grant, unavailable.
  • get_box: One of your cloud boxes: its state, its last 12 messages and its whole last reply. API: GET /v1/me/boxes/{box.id}, the rest of the inputs in the query. CLI: spacesheep boxes get <box>. Inputs: box | box_id (box name | box id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected, read-only ones too (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable.
  • stop_box: Stop a cloud box's current turn and drop the messages waiting behind it. The box and its files stay. API: POST /v1/me/boxes/{box.id}:stop, the rest of the inputs in a JSON body. CLI: spacesheep boxes stop <box>. Inputs: box | box_id (box name | box id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable.
  • update_box: Change a cloud box's YOLO mode, effort or model, from its next turn. Only what's given changes. API: PATCH /v1/me/boxes/{box.id}, the rest of the inputs in a JSON body. CLI: spacesheep boxes update <box> [flags]. Inputs: box | box_id (box name | box id, required), yolo (true or false), effort (one of low, medium, high, xhigh, max, default), model (text). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable. Old names, still answered: configure_box.
  • delete_box: Delete a cloud box: its container, files and disk. Its session and thread stay readable on /sessions. API: DELETE /v1/me/boxes/{box.id}, the rest of the inputs in the query. CLI: spacesheep boxes delete <box>. Inputs: box | box_id (box name | box id, required). Who can call it: a person in the browser; your API keys (the CLI, scripts); apps you connected (Claude, ChatGPT, Gemini, Grok); spacesheep's own agents. Refusals: forbidden, invalid, not_found, read_only_grant, unavailable. Old names, still answered: clear_box.

API

  • The calls above as HTTPS requests to https://spacesheep.dev/v1. OpenAPI 3.1: https://spacesheep.dev/v1/openapi.json. The server's own table: https://spacesheep.dev/v1/doors.json. Reference: https://spacesheep.dev/docs/api
  • Sign with Authorization: Bearer ss_… (an API key from https://spacesheep.dev/settings/api-keys#create) or an OAuth access token from a connected app; a browser signed in to spacesheep.dev sends its session.
  • Each {…} in a route is an input (a space as its uuid, @handle/slug or link, sent as one encoded part); the others go in the query (repeat a key for a list) or in a JSON object body, as each line says. A team, group, member, stack, design system, box, config group, MCP server or skill in a route is its id; to name one instead, send the name in the path and again as team, group, member, stack, design_system, box, config_group, mcp_server or skill in the query or body.
  • An answer is a JSON object (200). A refusal is {"code", "message", "violations": [{"field", "message"}]} with its code's HTTP status (the words under "Calls"); rate_limited also sends Retry-After, in seconds.
  • POST /v1/spaces (create_space). JSON body: title, emoji, description, visibility, listed, team, team_id, team_access, design_system, archived, official, files (required), label, client, slug, people, people_ids, org.
  • PATCH /v1/spaces/{space.id} (update_space). Path: space | space_id. JSON body: title, emoji, description, visibility, listed, team, team_id, team_access, design_system, archived, official.
  • POST /v1/spaces/{space.id}:share (share_space). Path: space | space_id. JSON body: people | people_ids (required), role.
  • DELETE /v1/spaces/{space.id} (delete_space). Path: space | space_id.
  • POST /v1/spaces/{space.id}/versions (create_version). Path: space | space_id. JSON body: files, label, parent, replacements, client, background, live_patch, drop_files.
  • POST /v1/spaces/{space.id}/versions/{version.id}:restore (restore_version). Path: space | space_id, version.
  • POST /v1/me/uploads (create_upload). JSON body: title, space, space_id, visibility, files.
  • POST /v1/spaces/{space.id}/highlights (create_highlight). Path: space | space_id. JSON body: highlights (required), color.
  • DELETE /v1/spaces/{space.id}/highlights/{highlight.id} (delete_highlight). Path: space | space_id, highlight.
  • GET /v1/me/captureLimits (get_capture_limits).
  • GET /v1/pageImages (list_page_images). Query: url (required), after, limit.
  • POST /v1/me/uploadFiles:import (import_upload_file). JSON body: url (required), path (required).
  • POST /v1/liveLinks (create_live_link). JSON body: title (required), emoji, description, status, visibility, team, team_id, client.
  • POST /v1/spaces/{space.id}/live:push (push_live). Path: space | space_id. JSON body: html, status, stream, edits, mode, stopped.
  • GET /v1/spaces/{space.id}/live/events (list_live_events). Path: space | space_id. Query: after, limit, wait_seconds.
  • GET /v1/spaces (list_spaces). Query: after, limit, include_archived.
  • GET /v1/spaces/{space.id} (get_space). Path: space | space_id.
  • GET /v1/spaces:search (search_spaces). Query: query (required), scope, after, limit.
  • GET /v1/spaces/{space.id}/versions (list_versions). Path: space | space_id. Query: after, limit.
  • GET /v1/spaces/{space.id}/files (list_files). Path: space | space_id. Query: version, after, limit.
  • GET /v1/spaces/{space.id}/files/{file} (get_file). Path: space | space_id, file. Query: version, offset, query.
  • GET /v1/me/publishAttempts (list_publish_attempts). Query: hours, only_refused, after, limit.
  • GET /v1/spaces/{space.id}/analytics (get_space_analytics). Path: space | space_id. Query: days.
  • GET /v1/spaces/{space.id}:screenshot (screenshot_space). Path: space | space_id. Query: path, full_page, width, height, section.
  • GET /v1/spaces/{space.id}:review (review_space). Path: space | space_id. Query: path.
  • GET /v1/spaces/{space.id}:export (export_space). Path: space | space_id. Query: paper, landscape.
  • GET /v1/spaces/{space.id}/comments (list_comments). Path: space | space_id. Query: after, limit.
  • POST /v1/spaces/{space.id}/comments (create_comment). Path: space | space_id. JSON body: body, block, quote, reply_to, resolved.
  • PATCH /v1/spaces/{space.id}/comments/{comment.id} (update_comment). Path: comment, space | space_id. JSON body: resolved.
  • POST /v1/spaces/{space.id}/comments/{comment.id}/reactions (create_reaction). Path: space | space_id, comment. JSON body: emoji (required).
  • GET /v1/spaces/{space.id}/discussionPosts (list_discussion_posts). Path: space | space_id. Query: topic, wait_seconds, after, limit, wait_after.
  • POST /v1/spaces/{space.id}/discussionPosts (create_discussion_post). Path: space | space_id. JSON body: text (required), title, reply_to, anchor, client.
  • PATCH /v1/spaces/{space.id}/discussionTopics/{discussion_topic.id} (update_discussion_topic). Path: discussion_topic, space | space_id. JSON body: outcome, resolved.
  • GET /v1/me/groups (list_groups). Query: group, group_id, after, limit.
  • POST /v1/me/groups (create_group). JSON body: title (required), add, remove, share, share_id, unshare, unshare_id.
  • PATCH /v1/me/groups/{group.id} (update_group). Path: group_id. JSON body: group (to name it: the same name in the path too), title, add, remove, share, share_id, unshare, unshare_id.
  • DELETE /v1/me/groups/{group.id} (delete_group). Path: group_id. Query: group (to name it: the same name in the path too).
  • GET /v1/spaces/{space.id}/signatureRequests (list_signature_requests). Path: space | space_id. Query: after, limit.
  • POST /v1/spaces/{space.id}/signatureRequests (create_signature_request). Path: space | space_id. JSON body: version, parties (required), note, supersedes.
  • POST /v1/spaces/{space.id}/signatureRequests/{signature_request.id}:withdraw (withdraw_signature_request). Path: space | space_id, signature_request.
  • GET /v1/me/referralLinks (list_referral_links). Query: after, limit.
  • POST /v1/me/referralLinks (create_referral_link). JSON body: target, label, archived, referral_link_id.
  • PATCH /v1/me/referralLinks/{referral_link.id} (update_referral_link). Path: referral_link. JSON body: archived.
  • GET /v1/me/designSystems (list_design_systems). Query: after, limit.
  • GET /v1/me/designSystems/{design_system.id} (get_design_system). Path: design_system_id. Query: design_system (to name it: the same name in the path too), version.
  • POST /v1/me/designSystems (create_design_system). JSON body: title (required), markdown (required).
  • PATCH /v1/me/designSystems/{design_system.id} (update_design_system). Path: design_system_id. JSON body: design_system (to name it: the same name in the path too), title, markdown.
  • DELETE /v1/me/designSystems/{design_system.id} (delete_design_system). Path: design_system_id. Query: design_system (to name it: the same name in the path too).
  • POST /v1/me/designSystems/{design_system.id}:share (share_design_system). Path: design_system_id. JSON body: design_system (to name it: the same name in the path too), teams, teams_ids.
  • GET /v1/me/spaceTransfers/{space_transfer.id} (get_space_transfer). Path: space_transfer.
  • GET /v1/me/spaceTransfers (list_space_transfers). Query: after, limit.
  • POST /v1/me/spaceTransfers (create_space_transfer). JSON body: space | space_id (required), recipient | recipient_id (required), request_id (required).
  • POST /v1/me/spaceTransfers/{space_transfer.id}:accept (accept_space_transfer). Path: space_transfer. JSON body: revision (required), terms (required).
  • POST /v1/me/spaceTransfers/{space_transfer.id}:decline (decline_space_transfer). Path: space_transfer. JSON body: revision (required).
  • POST /v1/me/spaceTransfers/{space_transfer.id}:cancel (cancel_space_transfer). Path: space_transfer. JSON body: revision (required).
  • POST /v1/spaces/{space.id}:copy (copy_space). Path: space | space_id. JSON body: request_id.
  • GET /v1/spaces/{space.id}/copyPermission (get_copy_permission). Path: space | space_id.
  • PATCH /v1/spaces/{space.id}/copyPermission (update_copy_permission). Path: space | space_id. JSON body: allowed, revision (required).
  • GET /v1/spaces/{space.id}/roomKey (get_room_key). Path: space | space_id.
  • PATCH /v1/spaces/{space.id}/roomKey (update_room_key). Path: space | space_id. JSON body: seats, hours, renew.
  • DELETE /v1/spaces/{space.id}/roomKey (delete_room_key). Path: space | space_id.
  • GET /v1/spaces/{space.id}/customDomain (get_custom_domain). Path: space | space_id.
  • GET /v1/me/customDomains (list_custom_domains). Query: after, limit.
  • PATCH /v1/spaces/{space.id}/customDomain (update_custom_domain). Path: space | space_id. JSON body: hostname.
  • DELETE /v1/spaces/{space.id}/customDomain (delete_custom_domain). Path: space | space_id.
  • POST /v1/spaces/{space.id}/publishSources:set (set_publish_source). Path: space | space_id. JSON body: publish_source (required).
  • GET /v1/me/stacks (list_stacks). Query: team, team_id, stack, stack_id, space, space_id, after, limit.
  • POST /v1/me/stacks (create_stack). JSON body: team, team_id, title (required), emoji, shelf, about, spaces, spaces_ids, also.
  • PATCH /v1/me/stacks/{stack.id} (update_stack). Path: stack_id. JSON body: stack (to name it: the same name in the path too), title, emoji, shelf, about, team, team_id.
  • DELETE /v1/me/stacks/{stack.id} (delete_stack). Path: stack_id. Query: stack (to name it: the same name in the path too), team, team_id.
  • POST /v1/me/stackEntries (create_stack_entry). JSON body: team, team_id, spaces | spaces_ids (required), stack, stack_id, how.
  • POST /v1/me/stacks:merge (merge_stacks). JSON body: team, team_id, from | from_id (required), into | into_id (required).
  • GET /v1/skills:search (search_skills). Query: query (required), source, level, after, limit, within_grants.
  • GET /v1/skills/{skill.id} (get_skill). Path: skill_id. Query: skill (to name it: the same name in the path too), view, path, version, offset.
  • GET /v1/skills (list_skills). Query: after, limit, team, team_id.
  • POST /v1/skills (create_skill). JSON body: files (required), slug, label, client.
  • PATCH /v1/skills/{skill.id} (update_skill). Path: skill_id. JSON body: skill (to name it: the same name in the path too), visibility, listed, team, team_id.
  • POST /v1/skills/{skill.id}/versions (create_skill_version). Path: skill_id. JSON body: skill (to name it: the same name in the path too), files (required), label, parent (required), client.
  • DELETE /v1/skills/{skill.id} (delete_skill). Path: skill_id. Query: skill (to name it: the same name in the path too).
  • POST /v1/skills:register (register_skill). JSON body: repo (required), check.
  • GET /v1/me/skill-repos (list_skill_repos).
  • GET /v1/me/installs (list_installs). Query: after, limit, team, team_id.
  • PATCH /v1/me/installs/{install.id} (update_install). Path: install. JSON body: mode, version, latest.
  • DELETE /v1/me/installs/{install.id} (delete_install). Path: install.
  • GET /v1/teams (list_teams). Query: after, limit.
  • PATCH /v1/teams/{team.id} (update_team). Path: team_id. JSON body: team (to name it: the same name in the path too), title, your_default, domains.
  • POST /v1/teams/{team.id}/members (create_team_member). Path: team_id. JSON body: team (to name it: the same name in the path too), member | member_id (required).
  • PATCH /v1/teams/{team.id}/members/{member.id} (update_team_member). Path: member_id, team_id. JSON body: member (to name it: the same name in the path too), team (to name it: the same name in the path too), role.
  • DELETE /v1/teams/{team.id}/members/{member.id} (delete_team_member). Path: team_id, member_id. Query: team (to name it: the same name in the path too), member (to name it: the same name in the path too).
  • POST /v1/teams (create_team). JSON body: title (required), your_default, domains, slug.
  • GET /v1/me/secrets (list_secrets). Query: team, team_id, config_group, config_group_id, after, limit.
  • PATCH /v1/me/secrets/{secret.id} (update_secret). Path: secret. JSON body: value, allowed_urls, team, team_id, config_group, config_group_id.
  • DELETE /v1/me/secrets/{secret.id} (delete_secret). Path: secret. Query: team, team_id, config_group, config_group_id.
  • GET /v1/me/configGroups (list_config_groups). Query: after, limit.
  • POST /v1/me/configGroups (create_config_group). JSON body: title (required), starter, invite, invite_ids, copy_secrets, unused_secrets, accepted.
  • PATCH /v1/me/configGroups/{config_group.id} (update_config_group). Path: config_group_id. JSON body: config_group (to name it: the same name in the path too), title, starter, invite, invite_ids, copy_secrets, unused_secrets, accepted.
  • DELETE /v1/me/configGroups/{config_group.id}/members/{member.id} (delete_config_group_member). Path: config_group_id, member_id. Query: config_group (to name it: the same name in the path too), member (to name it: the same name in the path too).
  • DELETE /v1/me/configGroups/{config_group.id} (delete_config_group). Path: config_group_id. Query: config_group (to name it: the same name in the path too).
  • GET /v1/me (get_account).
  • PATCH /v1/me (update_account). JSON body: handle, first_name, new_page_access, last_name, photo_key, photo_url, sounds_muted, sounds_off, language, format_sorting.
  • GET /v1/me/apiKeys (list_api_keys). Query: after, limit.
  • POST /v1/me/apiKeys (create_api_key). JSON body: label, scope.
  • POST /v1/me/apiKeys/{api_key.id}:revoke (revoke_api_key). Path: api_key.
  • GET /v1/me/agentModel (get_agent_model).
  • PATCH /v1/me/agentModel (update_agent_model). JSON body: provider, model, fast_model, premium_model.
  • POST /v1/me/codes:redeem (redeem_code). JSON body: code (required), team, team_id.
  • POST /v1/me/feedback (create_feedback). JSON body: message (required), category, tags, metadata, client_id (required), attachment.
  • GET /v1/me/backup (get_backup).
  • PATCH /v1/me/backup (update_backup). JSON body: enabled, hour_utc, run_now.
  • POST /v1/me/backup:disconnect (disconnect_backup).
  • GET /v1/me/channels (list_channels). Query: after, limit.
  • POST /v1/me/channels/{channel}:link (link_channel). Path: channel.
  • POST /v1/me/channels/{channel.id}:unlink (unlink_channel). Path: channel.
  • GET /v1/me/integrations (list_integrations). Query: after, limit.
  • POST /v1/me/integrations/{integration.id}:disconnect (disconnect_integration). Path: integration.
  • GET /v1/me/mcpServers (list_mcp_servers). Query: after, limit.
  • POST /v1/me/mcpServers (create_mcp_server). JSON body: url (required), title, team, team_id, actions.
  • DELETE /v1/me/mcpServers/{mcp_server.id} (delete_mcp_server). Path: mcp_server_id. Query: mcp_server (to name it: the same name in the path too), team, team_id.
  • POST /v1/me/integrations/{integration}:query (query_integration). Path: integration. JSON body: path (required), body, method, url, account.
  • POST /v1/me/agent:message (message_agent). JSON body: message (required), wait.
  • GET /v1/me/agent/replies/{agent_reply.id} (get_agent_reply). Path: agent_reply. Query: wait.
  • POST /v1/me/memories (create_memory). JSON body: content (required), kind, summary.
  • GET /v1/me/memories:recall (recall_memories). Query: query, after, limit.
  • DELETE /v1/me/memories/{memory.id} (delete_memory). Path: memory.
  • GET /v1/me/history:recall (recall_history). Query: query, from, to, limit.
  • GET /v1/me/photos (list_photos). Query: from, to, query, after, limit.
  • GET /v1/me/photos/{photo.id} (get_photo). Path: photo.
  • GET /v1/me/agentRuns (list_agent_runs). Query: hours, after, limit.
  • GET /v1/me/agentRuns/{agent_run.id} (get_agent_run). Path: agent_run.
  • GET /v1/me/streams (list_streams). Query: after, limit, prefix, points, health.
  • POST /v1/me/streams/{stream}:push (StreamService.PushStream). Path: stream. JSON body: value (required).
  • POST /v1/spaces/{space.id}:watch (watch_space). Path: space | space_id. JSON body: collection, note.
  • GET /v1/me/watches (list_watches). Query: after, limit.
  • DELETE /v1/me/watches/{watch.id} (delete_watch). Path: watch.
  • POST /v1/me/notifications:send (send_notification). JSON body: subject (required), message, space, space_id.
  • GET /v1/me/sessions (list_sessions). Query: after, limit, source, state, machine, since.
  • GET /v1/me/sessions/{session.id} (get_session). Path: session. Query: limit.
  • POST /v1/me/sessions/{session.id}:share (share_session). Path: session. JSON body: access, team, team_id, add, add_ids, remove, remove_ids, stop.
  • GET /v1/me/sessions:export (export_sessions). Query: since.
  • GET /v1/me/sessions/{session.id}/files (list_session_files). Path: session. Query: after, limit.
  • POST /v1/me/talkListeners (create_talk_listener). JSON body: session (required).
  • POST /v1/me/talkMessages (create_talk_message). JSON body: session (required), text | blocks (required), reply_to.
  • GET /v1/me/talkSettings (get_talk_settings).
  • PATCH /v1/me/talkSettings (update_talk_settings). JSON body: enabled.
  • POST /v1/me/boxes (create_box). JSON body: yolo, effort, model, prompt, agent, runs_on, repos.
  • POST /v1/me/boxes/{box.id}:message (message_box). Path: box_id. JSON body: box (to name it: the same name in the path too), prompt (required).
  • GET /v1/me/boxes (list_boxes). Query: after, limit.
  • GET /v1/me/boxes/{box.id} (get_box). Path: box_id. Query: box (to name it: the same name in the path too).
  • POST /v1/me/boxes/{box.id}:stop (stop_box). Path: box_id. JSON body: box (to name it: the same name in the path too).
  • PATCH /v1/me/boxes/{box.id} (update_box). Path: box_id. JSON body: box (to name it: the same name in the path too), yolo, effort, model.
  • DELETE /v1/me/boxes/{box.id} (delete_box). Path: box_id. Query: box (to name it: the same name in the path too).

CLI

  • npx -y spacesheep@latest <noun> <verb> (CLI 1.36 or newer) runs the calls above; the CLI reads them from https://spacesheep.dev/v1/doors.json. Reference: https://spacesheep.dev/docs/cli-commands
  • spacesheep <noun> <verb> --help lists a command's flags and refusals. A list flag takes commas, a JSON flag takes JSON (or - for stdin), a true-or-false flag needs no value; --json prints the answer as it came.
  • A refusal prints its message, ending with its word, and exits 1; a command line the CLI can't read exits 2, before anything is sent.
  • spacesheep spaces create <title> --files <list of files> [flags] (create_space). Flags: --emoji, --description, --visibility, --listed, --team, --team-id, --team-access, --design-system, --archived, --official, --files (JSON), --label, --client (JSON), --slug, --people, --people-ids, --org.
  • spacesheep spaces update <space> [flags] (update_space). Flags: --title, --emoji, --description, --visibility, --listed, --team, --team-id, --team-access, --design-system, --archived, --official.
  • spacesheep spaces share <space> (--people <list of person name> | --people-ids <list of person id>) [flags] (share_space). Flags: --people, --people-ids, --role.
  • spacesheep spaces delete <space> (delete_space).
  • spacesheep versions create (--space <space name> | --space-id <space id>) [flags] (create_version). Flags: --space, --space-id, --files (JSON), --label, --parent, --replacements (JSON), --client (JSON), --background, --live-patch (JSON), --drop-files.
  • spacesheep versions restore <version> (--space <space name> | --space-id <space id>) (restore_version). Flags: --space, --space-id.
  • spacesheep uploads create [flags] (create_upload). Flags: --title, --space, --space-id, --visibility, --files (JSON).
  • spacesheep highlights create (--space <space name> | --space-id <space id>) --highlights <list of highlights> [flags] (create_highlight). Flags: --space, --space-id, --highlights (JSON), --color.
  • spacesheep highlights delete <highlight> (--space <space name> | --space-id <space id>) (delete_highlight). Flags: --space, --space-id.
  • spacesheep capture-limits get (get_capture_limits).
  • spacesheep page-images list --url <text> [flags] (list_page_images). Flags: --url, --after, --limit.
  • spacesheep upload-files import --url <text> --path <text> (import_upload_file). Flags: --url, --path.
  • spacesheep live-links create <title> [flags] (create_live_link). Flags: --emoji, --description, --status, --visibility, --team, --team-id, --client (JSON).
  • spacesheep live push (--space <space name> | --space-id <space id>) [flags] (push_live). Flags: --space, --space-id, --html, --status, --stream (JSON), --edits (JSON), --mode, --stopped.
  • spacesheep live-events list (--space <space name> | --space-id <space id>) [flags] (list_live_events). Flags: --space, --space-id, --after, --limit, --wait-seconds.
  • spacesheep spaces list [flags] (list_spaces). Flags: --after, --limit, --include-archived.
  • spacesheep spaces get <space> (get_space).
  • spacesheep spaces search --query <text> [flags] (search_spaces). Flags: --query, --scope, --after, --limit.
  • spacesheep versions list (--space <space name> | --space-id <space id>) [flags] (list_versions). Flags: --space, --space-id, --after, --limit.
  • spacesheep files list (--space <space name> | --space-id <space id>) [flags] (list_files). Flags: --space, --space-id, --version, --after, --limit.
  • spacesheep files get <file> (--space <space name> | --space-id <space id>) [flags] (get_file). Flags: --space, --space-id, --version, --offset, --query.
  • spacesheep publish-attempts list [flags] (list_publish_attempts). Flags: --hours, --only-refused, --after, --limit.
  • spacesheep analytics get (--space <space name> | --space-id <space id>) [flags] (get_space_analytics). Flags: --space, --space-id, --days.
  • spacesheep spaces screenshot <space> [flags] (screenshot_space). Flags: --path, --full-page, --width, --height, --section.
  • spacesheep spaces review <space> [flags] (review_space). Flags: --path.
  • spacesheep spaces export <space> [flags] (export_space). Flags: --paper, --landscape.
  • spacesheep comments list (--space <space name> | --space-id <space id>) [flags] (list_comments). Flags: --space, --space-id, --after, --limit.
  • spacesheep comments create <body> (--space <space name> | --space-id <space id>) [flags] (create_comment). Flags: --space, --space-id, --block, --quote, --reply-to, --resolved.
  • spacesheep comments update <comment> (--space <space name> | --space-id <space id>) [flags] (update_comment). Flags: --space, --space-id, --resolved.
  • spacesheep reactions create (--space <space name> | --space-id <space id>) --comment <comment id> --emoji <text> (create_reaction). Flags: --space, --space-id, --comment, --emoji.
  • spacesheep discussion-posts list (--space <space name> | --space-id <space id>) [flags] (list_discussion_posts). Flags: --space, --space-id, --topic, --wait-seconds, --after, --limit, --wait-after.
  • spacesheep discussion-posts create <text> (--space <space name> | --space-id <space id>) [flags] (create_discussion_post). Flags: --space, --space-id, --title, --reply-to, --anchor, --client (JSON).
  • spacesheep discussion-topics update <discussion_topic> (--space <space name> | --space-id <space id>) [flags] (update_discussion_topic). Flags: --space, --space-id, --outcome, --resolved.
  • spacesheep groups list [flags] (list_groups). Flags: --group, --group-id, --after, --limit.
  • spacesheep groups create <title> [flags] (create_group). Flags: --add, --remove, --share, --share-id, --unshare, --unshare-id.
  • spacesheep groups update <group> [flags] (update_group). Flags: --title, --add, --remove, --share, --share-id, --unshare, --unshare-id.
  • spacesheep groups delete <group> (delete_group).
  • spacesheep signature-requests list (--space <space name> | --space-id <space id>) [flags] (list_signature_requests). Flags: --space, --space-id, --after, --limit.
  • spacesheep signature-requests create (--space <space name> | --space-id <space id>) --parties <list of signature parties> [flags] (create_signature_request). Flags: --space, --space-id, --version, --parties (JSON), --note, --supersedes.
  • spacesheep signature-requests withdraw <signature_request> (--space <space name> | --space-id <space id>) (withdraw_signature_request). Flags: --space, --space-id.
  • spacesheep referral-links list [flags] (list_referral_links). Flags: --after, --limit.
  • spacesheep referral-links create [flags] (create_referral_link). Flags: --target, --label, --archived, --referral-link-id.
  • spacesheep referral-links update <referral_link> [flags] (update_referral_link). Flags: --archived.
  • spacesheep design-systems list [flags] (list_design_systems). Flags: --after, --limit.
  • spacesheep design-systems get <design_system> [flags] (get_design_system). Flags: --version.
  • spacesheep design-systems create <title> --markdown <text> (create_design_system). Flags: --markdown.
  • spacesheep design-systems update <design_system> [flags] (update_design_system). Flags: --title, --markdown.
  • spacesheep design-systems delete <design_system> (delete_design_system).
  • spacesheep design-systems share <design_system> [flags] (share_design_system). Flags: --teams, --teams-ids.
  • spacesheep space-transfers get <space_transfer> (get_space_transfer).
  • spacesheep space-transfers list [flags] (list_space_transfers). Flags: --after, --limit.
  • spacesheep space-transfers create (--space <space name> | --space-id <space id>) (--recipient <person name> | --recipient-id <person id>) --request-id <text> (create_space_transfer). Flags: --space, --space-id, --recipient, --recipient-id, --request-id.
  • spacesheep space-transfers accept <space_transfer> --revision <integer> --terms <text> (accept_space_transfer). Flags: --revision, --terms.
  • spacesheep space-transfers decline <space_transfer> --revision <integer> (decline_space_transfer). Flags: --revision.
  • spacesheep space-transfers cancel <space_transfer> --revision <integer> (cancel_space_transfer). Flags: --revision.
  • spacesheep spaces copy <space> [flags] (copy_space). Flags: --request-id.
  • spacesheep copy-permission get (--space <space name> | --space-id <space id>) (get_copy_permission). Flags: --space, --space-id.
  • spacesheep copy-permission update (--space <space name> | --space-id <space id>) --revision <64-bit whole number> [flags] (update_copy_permission). Flags: --space, --space-id, --allowed, --revision.
  • spacesheep room-key get (--space <space name> | --space-id <space id>) (get_room_key). Flags: --space, --space-id.
  • spacesheep room-key update (--space <space name> | --space-id <space id>) [flags] (update_room_key). Flags: --space, --space-id, --seats, --hours, --renew.
  • spacesheep room-key delete (--space <space name> | --space-id <space id>) (delete_room_key). Flags: --space, --space-id.
  • spacesheep custom-domains get (--space <space name> | --space-id <space id>) (get_custom_domain). Flags: --space, --space-id.
  • spacesheep custom-domains list [flags] (list_custom_domains). Flags: --after, --limit.
  • spacesheep custom-domains update (--space <space name> | --space-id <space id>) [flags] (update_custom_domain). Flags: --space, --space-id, --hostname.
  • spacesheep custom-domains delete (--space <space name> | --space-id <space id>) (delete_custom_domain). Flags: --space, --space-id.
  • spacesheep publish-sources set <publish_source> (--space <space name> | --space-id <space id>) (set_publish_source). Flags: --space, --space-id.
  • spacesheep stacks list [flags] (list_stacks). Flags: --team, --team-id, --stack, --stack-id, --space, --space-id, --after, --limit.
  • spacesheep stacks create <title> [flags] (create_stack). Flags: --team, --team-id, --emoji, --shelf, --about, --spaces, --spaces-ids, --also.
  • spacesheep stacks update <stack> [flags] (update_stack). Flags: --title, --emoji, --shelf, --about, --team, --team-id.
  • spacesheep stacks delete <stack> [flags] (delete_stack). Flags: --team, --team-id.
  • spacesheep stack-entries create (--spaces <list of space name> | --spaces-ids <list of space id>) [flags] (create_stack_entry). Flags: --team, --team-id, --spaces, --spaces-ids, --stack, --stack-id, --how.
  • spacesheep stacks merge (--from <stack name> | --from-id <stack id>) (--into <stack name> | --into-id <stack id>) [flags] (merge_stacks). Flags: --team, --team-id, --from, --from-id, --into, --into-id.
  • spacesheep skills search --query <text> [flags] (search_skills). Flags: --query, --source, --level, --after, --limit, --within-grants.
  • spacesheep skills get <skill> [flags] (get_skill). Flags: --view, --path, --version, --offset.
  • spacesheep skills list [flags] (list_skills). Flags: --after, --limit, --team, --team-id.
  • spacesheep skills create --files <list of skill files> [flags] (create_skill). Flags: --files (JSON), --slug, --label, --client (JSON).
  • spacesheep skills update <skill> [flags] (update_skill). Flags: --visibility, --listed, --team, --team-id.
  • spacesheep skill-versions create (--skill <skill name> | --skill-id <skill id>) --files <list of skill files> --parent <integer> [flags] (create_skill_version). Flags: --skill, --skill-id, --files (JSON), --label, --parent, --client (JSON).
  • spacesheep skills delete <skill> (delete_skill).
  • spacesheep skills register --repo <text> [flags] (register_skill). Flags: --repo, --check.
  • spacesheep skill-repos list (list_skill_repos).
  • spacesheep installs list [flags] (list_installs). Flags: --after, --limit, --team, --team-id.
  • spacesheep installs update <install> [flags] (update_install). Flags: --mode, --version, --latest.
  • spacesheep installs delete <install> (delete_install).
  • spacesheep teams list [flags] (list_teams). Flags: --after, --limit.
  • spacesheep teams update <team> [flags] (update_team). Flags: --title, --your-default, --domains.
  • spacesheep team-members create (--team <team name> | --team-id <team id>) (--member <person name> | --member-id <person id>) (create_team_member). Flags: --team, --team-id, --member, --member-id.
  • spacesheep team-members update <member> (--team <team name> | --team-id <team id>) [flags] (update_team_member). Flags: --team, --team-id, --role.
  • spacesheep team-members delete <member> (--team <team name> | --team-id <team id>) (delete_team_member). Flags: --team, --team-id.
  • spacesheep teams create <title> [flags] (create_team). Flags: --your-default, --domains, --slug.
  • spacesheep secrets list [flags] (list_secrets). Flags: --team, --team-id, --config-group, --config-group-id, --after, --limit.
  • spacesheep secrets update <secret> [flags] (update_secret). Flags: --value, --allowed-urls, --team, --team-id, --config-group, --config-group-id.
  • spacesheep secrets delete <secret> [flags] (delete_secret). Flags: --team, --team-id, --config-group, --config-group-id.
  • spacesheep config-groups list [flags] (list_config_groups). Flags: --after, --limit.
  • spacesheep config-groups create <title> [flags] (create_config_group). Flags: --starter, --invite, --invite-ids, --copy-secrets, --unused-secrets, --accepted.
  • spacesheep config-groups update <config_group> [flags] (update_config_group). Flags: --title, --starter, --invite, --invite-ids, --copy-secrets, --unused-secrets, --accepted.
  • spacesheep config-group-members delete <member> (--config-group <config group name> | --config-group-id <config group id>) (delete_config_group_member). Flags: --config-group, --config-group-id.
  • spacesheep config-groups delete <config_group> (delete_config_group).
  • spacesheep me get (get_account).
  • spacesheep me update [flags] (update_account). Flags: --handle, --first-name, --new-page-access, --last-name, --photo-key, --photo-url, --sounds-muted, --sounds-off, --language, --format-sorting.
  • spacesheep api-keys list [flags] (list_api_keys). Flags: --after, --limit.
  • spacesheep api-keys create [flags] (create_api_key). Flags: --label, --scope.
  • spacesheep api-keys revoke <api_key> (revoke_api_key).
  • spacesheep agent-model get (get_agent_model).
  • spacesheep agent-model update [flags] (update_agent_model). Flags: --provider, --model, --fast-model, --premium-model.
  • spacesheep codes redeem <code> [flags] (redeem_code). Flags: --team, --team-id.
  • spacesheep feedback create <message> --client-id <text> [flags] (create_feedback). Flags: --category, --tags, --metadata (JSON), --client-id, --attachment (JSON).
  • spacesheep backup get (get_backup).
  • spacesheep backup update [flags] (update_backup). Flags: --enabled, --hour-utc, --run-now.
  • spacesheep backup disconnect (disconnect_backup).
  • spacesheep channels list [flags] (list_channels). Flags: --after, --limit.
  • spacesheep channels link <channel> (link_channel).
  • spacesheep channels unlink <channel> (unlink_channel).
  • spacesheep integrations list [flags] (list_integrations). Flags: --after, --limit.
  • spacesheep integrations disconnect <integration> (disconnect_integration).
  • spacesheep mcp-servers list [flags] (list_mcp_servers). Flags: --after, --limit.
  • spacesheep mcp-servers create --url <text> [flags] (create_mcp_server). Flags: --url, --title, --team, --team-id, --actions.
  • spacesheep mcp-servers delete <mcp_server> [flags] (delete_mcp_server). Flags: --team, --team-id.
  • spacesheep integrations query <integration> --path <text> [flags] (query_integration). Flags: --path, --body (JSON), --method, --url, --account.
  • spacesheep agent message --message <text> [flags] (message_agent). Flags: --message, --wait.
  • spacesheep agent-replies get <agent_reply> [flags] (get_agent_reply). Flags: --wait.
  • spacesheep memories create <content> [flags] (create_memory). Flags: --kind, --summary.
  • spacesheep memories recall [flags] (recall_memories). Flags: --query, --after, --limit.
  • spacesheep memories delete <memory> (delete_memory).
  • spacesheep history recall [flags] (recall_history). Flags: --query, --from, --to, --limit.
  • spacesheep photos list [flags] (list_photos). Flags: --from, --to, --query, --after, --limit.
  • spacesheep photos get <photo> (get_photo).
  • spacesheep agent-runs list [flags] (list_agent_runs). Flags: --hours, --after, --limit.
  • spacesheep agent-runs get <agent_run> (get_agent_run).
  • spacesheep streams list [flags] (list_streams). Flags: --after, --limit, --prefix, --points, --health.
  • spacesheep streams push <stream> --value <JSON value> (StreamService.PushStream). Flags: --value (JSON).
  • spacesheep spaces watch <space> [flags] (watch_space). Flags: --collection, --note.
  • spacesheep watches list [flags] (list_watches). Flags: --after, --limit.
  • spacesheep watches delete <watch> (delete_watch).
  • spacesheep notifications send --subject <text> [flags] (send_notification). Flags: --subject, --message, --space, --space-id.
  • spacesheep sessions list [flags] (list_sessions). Flags: --after, --limit, --source, --state, --machine, --since.
  • spacesheep sessions get <session> [flags] (get_session). Flags: --limit.
  • spacesheep sessions share <session> [flags] (share_session). Flags: --access, --team, --team-id, --add, --add-ids, --remove, --remove-ids, --stop.
  • spacesheep sessions export [flags] (export_sessions). Flags: --since.
  • spacesheep session-files list --session <session id> [flags] (list_session_files). Flags: --session, --after, --limit.
  • spacesheep talk-listeners create --session <session id> (create_talk_listener). Flags: --session.
  • spacesheep talk-messages create (--text <text> | --blocks <list of chat blocks>) --session <session id> [flags] (create_talk_message). Flags: --session, --text, --reply-to, --blocks (JSON).
  • spacesheep talk-settings get (get_talk_settings).
  • spacesheep talk-settings update [flags] (update_talk_settings). Flags: --enabled.
  • spacesheep boxes create [flags] (create_box). Flags: --yolo, --effort, --model, --prompt, --agent, --runs-on, --repos.
  • spacesheep boxes message <box> --prompt <text> (message_box). Flags: --prompt.
  • spacesheep boxes list [flags] (list_boxes). Flags: --after, --limit.
  • spacesheep boxes get <box> (get_box).
  • spacesheep boxes stop <box> (stop_box).
  • spacesheep boxes update <box> [flags] (update_box). Flags: --yolo, --effort, --model.
  • spacesheep boxes delete <box> (delete_box).

Deploy from a terminal or CI

  • Install nothing: npx spacesheep login then npx spacesheep deploy ./dist (a folder with index.html, or one .html file). Prints the URL. Run again to publish a new version.
  • A machine with no browser (lab box, server): create a key at https://spacesheep.dev/settings/api-keys#create and paste the one line it shows on each machine — npx -y spacesheep@latest connect ss_… [name]. Each machine gets its own key, named after its hostname.
  • Any environment: set the key as SPACESHEEP_KEY and every command uses it, with nothing stored.
  • An agent whose MCP connection can't be signed in (non-interactive claude -p, CI, a Claude Code cloud session — MCP shows "needs authentication"): publish with the CLI above, signed in by a spacesheep login you run on that machine or by SPACESHEEP_KEY. Cloud sessions: https://spacesheep.dev/setup.md (section "Cloud sessions").
  • Every MCP tool works from the CLI too: npx spacesheep delete <space> deletes one of your spaces, npx spacesheep tools lists the tools, and npx spacesheep call <tool> key=value key:=json runs any of them (pictures, such as a screenshot, are written to files). CLI 1.31 or newer; npx -y github:spacesheep-dev/spacesheep-cli is always the newest.
  • Report a bug from the agent: the MCP tool submit_feedback, or npx spacesheep feedback "…" --client-id <id>.
  • GitHub Actions: add an API key from https://spacesheep.dev/settings/api-keys#create as the SPACESHEEP_KEY secret, then - uses: spacesheep-dev/spacesheep-cli@v1 with dir: dist and key: ${{ secrets.SPACESHEEP_KEY }}.
  • Source: https://github.com/spacesheep-dev/spacesheep-cli (MIT). npm package: spacesheep.

Examples

  • Public AI artifact gallery: public spaces picked by the spacesheep team from those their creators show on their profiles. Sharing publicly never puts a space there by itself, and a creator keeps one out from its Share menu.

Chrome extension (for a person with the page open)

  • A person with anything open in Chrome (a page, a picture, a PDF, a video, a document) publishes it with the spacesheep Chrome extension: one click gives it its own link; a second click updates the same space. It also moves files and whole folders from their computer. Install and what it reads: https://spacesheep.dev/extension (source: https://github.com/spacesheep-dev/spacesheep-extension). Point a person there when they have a file in hand and no agent connected.

AGI-pilled signing: document signing for contracts, NDAs and offer letters an agent wrote

  • An agent that wrote an agreement publishes it as a page, then calls the MCP tool request_signatures({ uuid, parties: [{ email, name, role, on_behalf_of }], note }). Each party gets an email, signs in with that address and signs on the page with a typed name after a consent to sign electronically. The agent can send and track a request; it can never sign for anyone.
  • Mark each party's block in the page with data-ss-sign="<role>" (e.g. company, contractor) so their name and the date are drawn into it when they sign. signing_status says who has signed and links the certificate; withdraw_signature_request closes an open request; a new request with supersedes amends a signed one.
  • What is recorded: the SHA-256 of the signed version's files, each signer's typed name, consent, time, IP address and browser, each sealed with Ed25519 and verifiable against https://spacesheep.dev/.well-known/spacesheep-signing.json. Every party gets the signed version and its certificate as PDFs when the last one signs. Built around the US ESIGN Act and UETA.
  • Sending is a beta on every plan: a free account sends 15 requests in any 30 days once it is past its first week; Pro, Max and Team have no monthly limit (20 requests an hour, 100 a day). Signing is free on any plan. How it works, and how it compares with DocuSign, Dropbox Sign and PandaDoc: https://spacesheep.dev/signing

YC startups: a year of Team free, 15 members

  • An active YC founder gets their company a year of Team for up to 15 members; each member has everything Max has (signature requests, Speak, the agent on spacesheep's model key, every Pro limit). The agent claims it: get the founder's YC verification link (redeem spacesheep's deal on Bookface, e.g. through YC's MCP deal.redeem, which returns https://deals.ycombinator.com/redemptions/verify/…; or their https://www.ycombinator.com/verify/… founder link) and pass it as code to the MCP tool redeem_code (with org to put it on a team they own; otherwise it goes on their team named after the company, or a new one). It is checked with YC and must belong to the signed-in account's email or company domain; one claim per YC company. A person can paste it at https://spacesheep.dev/yc instead.

Streams (live data into a space)

  • Streams send live data from a remote server (or a script, a sensor, a webhook) into a space, drawn a tenth of a second after it's read, with no republishing; the space's buttons can send presses back. Sending needs Max; anyone who can open the space can watch. How: https://spacesheep.dev/streams (agents: the skill's streams section, and the MCP tool streams).
  • A space that isn't streaming: https://spacesheep.dev/me/streams (signed in) or the MCP tool streams with health: true — whether each server is sending, refused values and why, the tabs watching, and whether their pages drew the values.
  • Live video (stream a camera into a space): publish a page with <meta name="ss-streams" content="lab/cam-1"> and <video data-ss-stream="lab/cam-1">, then open https://spacesheep.dev/film/lab/cam-1?space= in a browser on the phone or computer that films (camera, screen or test pattern, then Go live) — that is live video, about 30 frames a second, and always the first answer. Only with no browser (a headless server, an IP camera): npx -y github:spacesheep-dev/spacesheep-cli stream lab/cam-1 --camera --space <page link>, which sends 3 still pictures a second, not live video — say so. Filming needs Pro (1 camera at once) or Max/Team (5 at once); signed-in people who can open the page watch; nothing is recorded. Details: the streams section of the skill.

Email the person (a form's responses, a long task's end)

  • A page never sends email. To email the signed-in person when someone submits a page's form, RSVP, booking or poll, call the MCP tool watch_space with the page: the first response is emailed at once, then at most one digest an hour, with a count and a link (never what anyone wrote). For "email me when you're done", call notify_me with a subject and the page as space.
  • Both mail only the account's own address (there is no recipient), at most 3 every 10 minutes and 10 a day per person, and every email has a one-click stop. list_watches and stop_watch manage watches.

Reading a space link

  • A URL like https://spacesheep.dev/@handle/slug is a space. Public spaces are viewable anonymously; private ones need sign-in.
  • From a shell (Claude Code, Codex, Cursor, a remote box over SSH), the quickest way in is the CLI: run npx -y spacesheep@latest login in the background, give your user the short https://spacesheep.dev/cli?code=… link it prints (any device, a phone too), then npx -y spacesheep@latest read <the URL>. Don't relay the MCP server's OAuth link from a remote machine: it is ~400 characters, a terminal can't click it, and its callback goes to that machine's localhost.
  • With no shell (ChatGPT, claude.ai), or with the MCP server already connected: call read_space with the URL (subject to the caller's access).

Reading a second account's pages (read-only key)

  • A machine that publishes as one account can read another account's pages through a read-only key: the person makes it in that account (Settings → API keys → Read only, or create_api_key with scope "read") and runs claude mcp add --scope user --transport http spacesheep-<handle> https://mcp.spacesheep.dev/mcp --header 'Authorization: Bearer ss_…' on the machine.
  • That connection lists only list_files, get_file, search_spaces, list_spaces, list_versions and list_comments, and refuses every write. Publish through the machine's own spacesheep connection.

Channels (the same agent, reachable from chat apps)

  • Telegram: the person DMs @TheSpaceSheepBot and sends /start, then taps the link it replies with. Or a one-time code from https://spacesheep.dev/settings/telegram, sent as /start . In a group, /bind ties the group to an account or org. An agent cannot do the tap; relay these steps to the user. Details: https://spacesheep.dev/docs#telegram
  • Slack: an org owner connects the workspace from the org's Settings page on https://.spacesheep.dev/settings.
  • Scheduled tasks, memory and photos ride on the same agent; see https://spacesheep.dev/docs.

Docs