@voteship/mcp-server

MCP (Model Context Protocol) server for VoteShip — manage feature requests, votes, roadmaps, and AI workflows from any MCP-compatible client.

Works without an API key — public tools let any agent submit feedback, browse boards, vote, and comment.

Installation

Claude Code

Add to your project's .mcp.json:

{ "mcpServers": { "voteship": { "command": "npx", "args": ["@voteship/mcp-server"], "env": { "VOTESHIP_API_KEY": "sk_..." } } } }

Public Mode (No API Key)

For agents that just need to submit feedback or browse boards:

{ "mcpServers": { "voteship": { "command": "npx", "args": ["@voteship/mcp-server"], "env": { "VOTESHIP_PROJECT_SLUG": "my-project" } } } }

No VOTESHIP_API_KEY needed. The server starts in public mode with 4 tools for interacting with any public VoteShip board.

Cursor / Windsurf

Add to your MCP settings:

Find your API key in VoteShip → Settings → Share & Embed.

Public Tools (4) — No API Key Required

Tool	Description
submit_feature_request	Submit feedback to any public VoteShip board. Post created as PENDING (requires board owner approval).
browse_board	View approved feature requests on a public board, sorted by votes
upvote_post	Vote/unvote on a feature request (uses deterministic anonymous ID)
add_public_comment	Add a comment to a feature request

Admin Tools (25) — Requires API Key

Tool	Description
list_posts	List feature requests with optional filters
get_post	Get a single post with votes, comments, tags
create_post	Create a new feature request
update_post	Update a post's title, description, status, or tags
delete_post	Delete a feature request
search_similar	Find similar posts using AI semantic search
add_vote	Vote on a feature request
get_voters	List who voted on a post
add_comment	Add a comment to a post
get_comments	List comments on a post
delete_comment	Delete a comment from a post
list_tags	List all available tags
create_tag	Create a new tag
list_users	List board users
update_user	Update a board user's name, email, or spend value
get_roadmap	Get the product roadmap grouped by status
get_analytics	Get analytics summary for a time period
list_releases	List published changelog releases
create_release	Create a changelog release
submit_feedback	Submit unstructured text as a feature request (AI processes it)
triage_inbox	AI-powered triage of unreviewed posts
get_summary	AI-generated summary of recent feedback
plan_sprint	AI-suggested sprint based on votes and themes

Resources (5)

URI	Description
voteship://project/overview	Project info and summary stats
voteship://project/board	Full board state with all posts
voteship://project/roadmap	Public roadmap view
voteship://project/changelog	Published releases
voteship://project/analytics	Analytics snapshot

Prompts (5)

Name	Description
triage_inbox	Review and categorize unprocessed feature requests
sprint_planning	Suggest what to build next based on data
generate_changelog	Draft release notes from recently completed posts
feedback_summary	Summarize feedback trends and highlights
submit_feedback	Guide for submitting a well-structured feature request (works in public mode)

Example Usage

Public mode (any agent):

> Submit a feature request to the acme-app board: "Add dark mode support"

Admin mode (with API key):

> Triage my VoteShip inbox and suggest what to build this sprint

The agent will:

1. Fetch unreviewed posts and analyze them
2. Flag duplicates and suggest statuses
3. Recommend a sprint plan based on votes and themes
4. Generate changelog drafts for completed features

Environment Variables

Variable	Required	Description
VOTESHIP_API_KEY	No	API secret key (sk_...) for full admin access. Omit for public-only mode.
VOTESHIP_PROJECT_SLUG	No	Default project slug for public tools (e.g. my-app)
VOTESHIP_API_URL	No	Custom API URL (default: https://app.voteship.app)

Development

Testing

Run tests

pnpm test

Run with coverage

pnpm test:coverage

The MCP server has 82 tests covering:

• Client HTTP layer — URL construction, headers, auth, error handling for all 25+ methods
• Server registration — All 29 tools, 5 resources, 5 prompts verified
• Tool callbacks — Success/error paths, public tool slug resolution
• Resource callbacks — Data formatting and response structure

Coverage thresholds are enforced: 55% lines, 40% branches, 70% functions, 55% statements.

License

MIT