Your agent can query this map

One command connects Claude Code, Cursor, Codex — any MCP client — to 366 curated repos and every human-reviewed relationship, with the why written down.

claude mcp add -s user stackmap -- npx -y @ship-with-ai/stackmap-mcp

{ "mcpServers": { "stackmap": { "command": "npx", "args": ["-y", "@ship-with-ai/stackmap-mcp"] } } }

[mcp_servers.stackmap]
command = "npx"
args = ["-y", "@ship-with-ai/stackmap-mcp"]

Claude Code: -s user makes the server available in every project — the default scope registers it only for the directory you ran the command in. Start a fresh session after installing; /mcp shows what’s loaded. Saying “StackMap” in your question routes the agent straight to the tools.

What it answers

• search_catalog — “What’s on the StackMap for local voice transcription?”
• get_repo — “Tell me about langgraph — including when NOT to use it.”
• get_alternatives — “What replaces crewai, and when do I pick which?”
• continue_stack — “I use langgraph + chroma — what pairs well next?”
• build_stack — “Build me a starter stack for a local-first RAG pipeline over PDFs.”

Every answer opens with the catalog version (currently v2026.09.07) and quotes the curator’s written why — including when not to use a tool.

Nothing to trust blindly

• No auth, no state, no telemetry — the server reads this site’s public JSON API and nothing else.
• Public source — Ship-With-AI/stackmap-mcp, ~150 lines, CI-tested, published to npm with provenance.
• Human-reviewed data — nothing on the map auto-publishes; the refusal and dismissal ledgers are public.

Prefer raw JSON? /api/catalog.json · per-repo at /api/repos/<slug>.json · llms.txt

Want this rigor applied to your team’s stack? We do paid stack audits →