@mitchens84/sitruna-clickup-server
v0.1.0
Published
Sitruna-custom ClickUp MCP — 5 enhanced read tools (allowlist-stripped, v3 chat-reply attribution) + 27 write pass-throughs + 1 diagnostic. ~86% context reduction vs stock claude_ai_ClickUp. Supports stdio (local) + HTTP (remote) transports.
Maintainers
Readme
sitruna-clickup-server
Sitruna-custom ClickUp MCP. 5 enhanced read tools (allowlist-stripped, v3 chat-reply auto-fetch) + 27 write pass-throughs. Complements stock claude_ai_ClickUp (writes) rather than replacing it.
Status: scaffold (260424). Tool implementations land in Phase 3.1–3.3 per
99-PLANS/260422-7A-SITRUNA-CLICKUP_MCP_BUILD-PLAN.md.
Why
Stock ClickUp MCP returns ~14 KB per task. A get_tasks call for 100 tasks returns ~925 KB — larger than Claude's entire 200 K context window. This fork applies an allowlist (17 fields for tasks, 4 for comments, 3 for chat replies) before data reaches Claude. Measured reduction: ~90 % per task, 75–79 % for comments and chat threads.
Install (local dev)
pnpm install
pnpm buildConfigure
Copy .env.example → .env (or set the same vars in your MCP client's env block):
| Var | Required | Default | Purpose |
|---|---|---|---|
| CLICKUP_API_TOKEN | yes | — | ClickUp personal token (pk_...) |
| CLICKUP_AUTHOR_SKIP_LIST | no | Lemon,Agent 1 | Comma-separated usernames filtered out of comment tools |
| LOG_LEVEL | no | info | pino log level |
| LOG_FILE | no | ~/Library/Logs/sitruna-clickup-server.log | pino sink |
Register
Add to ~/.claude.json under mcpServers:
"sitruna-clickup-v2": {
"command": "node",
"args": ["/absolute/path/to/sitruna-clickup-server/build/index.js"],
"env": { "CLICKUP_API_TOKEN": "pk_..." }
}Cutover from sitruna-clickup-v2 → sitruna-clickup happens at Phase 6 DEPLOY once parity tests pass.
Benchmark
pnpm benchmarkRuns the stripper against tests/fixtures/*.json and emits reduction % per object type. Gate: ≥75 % per type (Phase 4.1).
Inspector
pnpm inspectorOpens localhost:5173 with tools/list, tools/call, etc. — the MCP-SDK visual debugger.
Tools (planned)
Reads (Phase 3.1)
| Tool | API | Input |
|---|---|---|
| get_tasks | v2 /list/{list_id}/task | list_id, 15 filter params |
| get_task_details | v2 /task/{task_id} | task_id |
| get_task_comments | v2 /task/{task_id}/comment | task_id, pagination |
| get_chat_view_comments | v2 + v3 | view_id — v3 replies auto-fetched |
| get_chat_message_replies | v3 | workspace_id, message_id |
Writes (Phase 3.2)
27 tools, pass-through from stock claude_ai_ClickUp schemas. Identical semantics.
References
- Execution plan:
99-PLANS/260422-7A-SITRUNA-CLICKUP_MCP_BUILD-PLAN.md - Lifecycle standard:
00-MASTER/03-STANDARDS/STD-MCP_LIFECYCLE.md - Projection rules:
00-MASTER/03-STANDARDS/STD-MCP_PROJECTION.md - Error / logging:
00-MASTER/03-STANDARDS/STD-MCP_ERROR_PROTOCOL.md - Design ref:
7A-PROJECTS-SITRUNA-MCP/00-SHARED/260416-MCP_DESIGN-BEFORE_AFTER_REFERENCE.md
