PinAppAI

@pinappai/mcp · npm

Your reviewers’ words, inside Claude Code.

Install one MCP server. Your AI coding tool reads change requests, finds patterns, and proposes the fix — without leaving the editor. Works in Claude Code, Cursor, Continue, Codex, and Claude Desktop.

Get an API key → View on npm

Why MCP

Stop copy-pasting feedback into your editor

Without MCP, the loop is: open the dashboard, read the queue, copy the markdown, paste into Claude Code, ask it to fix. With MCP, you stay in the editor — the agent reads the queue itself, finds the patterns across reviewers, and proposes diffs you actually want to apply.

Install

Two commands. Any AI client.

`install` registers the MCP server in every AI client config it finds on your machine. `login` opens your browser for sign-in, picks a workspace if you have multiple, and saves a per-machine API key locally. No copy-paste of long JSON. No per-host snippets to keep in sync.

npx @pinappai/mcp install
npx @pinappai/mcp login

Auto-detected on:

  • Claude Code
  • Cursor
  • Claude Desktop
  • Codex CLI
  • Continue

Need a CI / headless install, or want the manual config? See the manual install →

Tools reference

32 admin-parity tools — read your state, drive your workflow

Your AI client calls these on your behalf. The 6 read tools below are what you’ll touch most often; the other 26 cover project / workspace / member / reviewer / CR lifecycle — gated by API-key role.

Tool Purpose
list_projects List PinAppAI projects this API key can access
get_review_summary Cheap one-call summary of change-request state for a project
list_change_requests Paginated change-request list with filters (status, page, reviewer)
get_change_request Full CR detail with related[] siblings + optional inline screenshot
analyze_patterns Server-side groupings + heuristic clusters + optional AI clustering
get_screenshot Inline screenshot as an MCP image content block

Slash commands

Full workflows in one command

Slash commands ship with the MCP — no extra setup. Just type and your AI client runs the workflow.

  • /pinappai:setup-project

    Walk through workspace + project + widget snippet + reviewers from scratch

  • /pinappai:fix-changes

    Triage queued change requests, propose code edits in your repo

  • /pinappai:apply-decisions

    Apply queued reviewer decisions (rollbacks + change-requests) to source — one batch, marker-aware

  • /pinappai:generate-changes-page

    Regenerate the /changes/ review page since the last applied batch

  • /pinappai:remove

    Strip widget + /changes/ + helpers when you ship; preserves .pinappai/ for re-install

5 more in /docs/mcp/ — analyze, summarize, audit-review, triage-and-resolve, auth-help.

FAQ

Common questions

What is MCP?

Model Context Protocol — an open standard so AI clients (Claude Code, Cursor, Continue, etc.) can talk to external data sources through a uniform interface. Read more at modelcontextprotocol.io.

Do I need to be a developer?

No. The slash commands work for anyone using Claude Code or Cursor — type /pinappai:analyze and the AI does the rest.

Will it edit my code without asking?

No. Your AI client always confirms before applying any source edit. Every applies-to-source slash command (fix-changes, apply-decisions, remove) ends with an a/b/c/d picker — new branch, commit current, stage only, or skip — so commits never happen on autopilot. The MCP itself can mutate PinAppAI state (projects, members, CRs) when you ask it to, but it never touches your repo files directly.

Is it open source?

Yes — MIT licensed. Published as @pinappai/mcp on npm; you can read every line via npm view or in your node_modules after install.

What does it cost?

Free with any PinAppAI plan, including Free. The MCP just reads what you’ve already paid for.

Ready to bring your reviews into your editor?

Mint an API key, paste the install snippet, run a slash command. About 3 minutes start to finish.

Get started — free → Read on npm