Borker
Draft, write and schedule social posts in your own voice for X, LinkedIn, Bluesky, Farcaster and Paragraph, each one waiting for your approval.
Servidor MCP hospedado
npx add-mcp 'https://borker.xyz/api/v1/mcp'Instala no Claude Code, Codex, Cursor e outros
Documentação
One call, one draft
Your agent sends a draft. It waits for you with every other post.
Your agent sends a draft
API POST /api/v1/content MCP create_content_draft
{"channelId": "7d3f2c1a-...","body": "Three weeks of wind tables said wait."}
It arrives as a draft
PostsReviewed like any other post.
12:40Needs you
PersonalVia agent
Text your agent wrote costs no generation.
More than a way to post
A plain posting tool gives your agent a way to publish, and nothing more. What goes out is whatever the agent wrote, whenever it decided to post. Borker adds three things around the posting.
Your voice
Your agent reads the facts Borker may state about you, your rules for how you sound and the subjects you stay away from. Ask Borker to write the post and it writes it in the voice of the channel it goes to, yours or your brand's.
A planned week
Your agent sees how often each channel posts and what is planned for the next 14 days. A draft it makes takes the channel's next open time slot. Once approved, it counts toward that week, and Borker plans one fewer post of its own.
Your review
Every post your agent makes lands in the Review column of your Command Center, the page in Borker that shows your week. It is labeled Via agent and goes through the same checks as every other post. Even if you let Borker publish its own posts without you, it never publishes your agent's. An agent's post waits until you approve it, or until your agent schedules it with a key you gave write access.
Connect in one line
- In Borker, open Settings, then Keys, and create an agent key. Pick Write to let your agent draft and schedule, or Read to let it look only. Copy the key right away: Borker shows it only once.
- Add Borker to Claude Code:
claude mcp add --transport http borker https://borker.xyz/api/v1/mcp \ --header "Authorization: Bearer <your agent key>" - Ask Claude to list your Borker channels. Every draft has to name the channel it is for, and this gives Claude the list.
Claude Desktop and Cowork connect through their config file. Any other client that speaks MCP over streamable HTTP and can send a header connects directly. The Add custom connector dialog in the Claude apps does not work with Borker yet, because it has no field for an agent key.
Every client, step by step, in the MCP docs
What you can ask
- “List my Borker channels.”
Your connected channels come back with their ids, and whether each one speaks as you or as your brand.
list_channels - “What do I have planned for the next two weeks?”
Your agent reads how often each channel posts and every planned post for the next 14 days, written or not yet.
get_schedule - “Read my brand voice, then draft a LinkedIn post about today's release.”
Your agent writes the text from your voice profile and sends it as a draft for your LinkedIn channel. A draft your agent writes itself does not count against your plan's monthly posts. It waits in Review, timed for the channel's next open slot.
get_brandcreate_content_draft - “Have Borker write a post about the new onboarding for X and Bluesky.”
Borker writes one post per channel, each in the voice of its channel, and both wait in Review. Each one counts as a post from your plan's monthly allowance, unless you bring your own Anthropic key.
generate_content - “Show me the drafts waiting for review and tighten the one that sounds like AI.”
Your agent lists your drafts, each with the phrases Borker flags as sounding like AI, and rewrites the text. Borker checks the new text again.
list_contentupdate_content - “Schedule the launch post for Thursday at 9.”
The post is approved and queued for that time. If its text still has phrasing that sounds like AI, Borker asks for an explicit yes first.
schedule_content
What your agent can reach
- One workspace. A key belongs to the workspace it was made in, everything Borker holds for one brand, and never sees another one.
- Only what its scope, Read or Write, allows. A Read key can look at your channels, voice, plan, stats and posts. A Write key can also draft, have Borker write, edit, approve, schedule and delete posts that are not published. An agent cannot manage keys over MCP.
- Only what its maker may do. A key acts as the person who made it: a key made by a team member with view-only access can read and never write.
- Until you revoke it. Revoke a key in Settings and its next request is refused, with no grace period.
Scopes and revoking, in the agent key docs Every tool and endpoint, in the API reference
Questions
Yes, through Borker. Connect your LinkedIn profile or company page in Borker, then connect Claude to Borker's MCP server with an agent key. Claude can draft a LinkedIn post, have Borker write one in your voice, and schedule it. Borker publishes it, and Claude never holds your LinkedIn login.
No. Your agent holds one agent key, not your logins. You connect your channels in Borker yourself, and the agent never sees those connections. The key works only in your workspace, and you can revoke it in Settings at any moment.
Only with a Write key. A post your agent makes always arrives as a draft waiting for your approval, and Borker never publishes it on its own, even for a voice you let publish without you. A Write key can also approve and schedule a post, or publish it now, the way you could. Phrasing that sounds like AI needs an explicit yes first. A Read key can look but never write.
Yes, to every channel you connect in Borker. Claude asks Borker for your channels, and each draft names the one it is for. Each channel speaks as you or as your brand.
Text your agent writes and sends as a draft is free. When Borker writes the post, it counts as one post from your plan's monthly allowance for each channel, unless you bring your own Anthropic key. Agent keys come with every plan.
Building with Claude Code every day? Borker can turn your coding sessions into posts with a build log.