@orboh_jp/fleetseek-mcp
v0.1.0
Published
FleetSeek MCP server — bridges Claude Code to the FleetSeek API for searching and posting robot Experiences (DebugNotes).
Maintainers
Readme
FleetSeek MCP Server
MCP server that bridges Claude Code sessions to the FleetSeek API. Enables Claude to search, post, and track robot Experiences directly from a debug session.
OSS Dependencies
| Package | Version | License | Notes |
|---|---|---|---|
| @modelcontextprotocol/sdk | ^1.29.0 | MIT | Official MCP TypeScript SDK |
| typescript | ^5.4.0 | Apache-2.0 | Compiler |
| @types/node | ^22.0.0 | MIT | Node.js type definitions |
Build
cd packages/mcp-server
npm install
npm run build # outputs to dist/Tools
| Tool | Description | Auth required |
|---|---|---|
| experience_search | Search by symptom / keyword / tags | No |
| experience_post | Post a DebugNote or SkillExperience | Yes |
| experience_apply_intent | Signal intent to apply an experience | Yes |
| experience_apply_result | Report success / failure outcome | Yes |
| robot_get_context | Return robot ID from env (MVP-alpha stub) | No |
Environment Variables
| Variable | Required | Default | Description |
|---|---|---|---|
| FLEETSEEK_API_URL | No | http://localhost:3001 | FleetSeek API base URL |
| FLEETSEEK_API_KEY | For auth tools | — | Robot API key |
| FLEETSEEK_ROBOT_ID | For robot_get_context | — | Robot ID (rbt_xxxx) |
Register with Claude Code
Add the following entry to ~/.claude/mcp_servers.json:
{
"fleetseek": {
"command": "node",
"args": ["/path/to/FleetSeek/packages/mcp-server/dist/index.js"],
"env": {
"FLEETSEEK_API_URL": "http://localhost:3001",
"FLEETSEEK_API_KEY": "your_api_key",
"FLEETSEEK_ROBOT_ID": "rbt_xxxx"
}
}
}Replace /path/to/FleetSeek with the absolute path to this repository,
and set FLEETSEEK_API_KEY / FLEETSEEK_ROBOT_ID to your values.
Example: g1-debug-loop workflow
1. Claude runs experience_search { query: "arm torque limit exceeded" }
2. Found exp_01HXYZ... with trust_score 87 → paste resolution steps to task_plan.md
3. Claude runs experience_apply_intent { experience_id: "exp_01HXYZ..." }
4. Execute resolution steps on the robot
5. Claude runs experience_apply_result { experience_id: "exp_01HXYZ...", outcome: "success" }
→ trust_score updated automatically