MCP

/mcp — connector health-check & discovery

🧒 When reporting to Anton end with a child-simple "In plain words" recap. (memory eli5-always)

Two jobs: (1) health — is each MCP server alive & authorized? (2) discover — is there an MCP for a need Anton has? Full rules + the Telegram landmine = memory [[mcp-health-check]].

health (READ-ONLY, safe order)

1. In-session tool check FIRST (zero-risk) — if a server's tools are present in the session, call ONE cheap read tool to confirm auth + liveness:
   • Telegram → mcp__telegram__get_me (expect Tony / @work_acct_a).
   • WhatsApp → mcp__whatsapp__get_my_profile. n8n → mcp__n8n__n8n_health_check. Google Drive/Calendar → a list call. MongoDB → list-databases. This uses the ALREADY-RUNNING server — no second client.
2. Only if a server's tools are ABSENT → it didn't load. Diagnose registration ($USERPROFILE/.claude.json mcpServers) + processes, then reconnect.
3. ⚠️ TELEGRAM LANDMINE: do NOT run claude mcp list or spawn a Telethon client as a casual check — a 2nd client on the same StringSession triggers AUTH_KEY_DUPLICATED and logs the account out. Prefer in-session get_me. Use claude mcp list only when the server clearly isn't loaded.
4. Zombie/leak watch (Windows): stdio MCPs launched via uv run/npx can orphan → prefer the direct interpreter as command; the "Telegram MCP Janitor" task (every 15 min) kills true orphans (no live claude.exe ancestor). Don't guess a hung process — verify with facts (alive? progress mtime? elapsed?).

discover

Use the registry tools: mcp__mcp-registry__suggest_connectors (for a stated need), mcp__mcp-registry__search_mcp_registry, mcp__mcp-registry__list_connectors (what's already available). Rank by: does it solve a REAL recurring need? subscription/local (free) over paid ([[prefer-included-limits-before-paid-api]])? does a stdio server risk Windows orphaning? Propose, don't auto-install — wiring a new server = Anton's call.

Safety

• Health = read-only, run anytime (this is the standing pre-flight before Telegram/integration work).
• Reconnecting, editing .claude.json, installing a new server = pause + confirm.

Output

Per-server 🟢/🔴 (alive + authed) + any down one with its fix command; for discover, a short ranked shortlist. Then 🧒 recap. Setup detail: [[telegram-mcp-connector]], [[whatsapp-mcp-integration]], [[n8n-stack]].

Like this skill? It is one of 100 in second-brain-starter-kit: the second brain we built for ourselves and run every day at Palo Alto AI Research Lab. Install the whole set with npx skills add tonydzi/second-brain-starter-kit. Everything is open source and free, so take what you need.

Flagships worth a look on their own: secondop-panel (a second opinion from a panel of external models), claude-memory-tidy (stop your agent's memory from rotting), telegram-mcp-kit (your own Telegram over MCP in about 15 minutes).

Author: Anton Dziatkovskii, Palo Alto AI Research Lab. Telegram @tonydzi - WhatsApp +1 341 222 9178 - X @Tony_Stef_

Engineers: want to test-drive this setup? Message me. I hand out free starter seeds to engineers who test and report back, and custom skill requests are welcome.