Yunxiao MCP Server
中文文档
Let your AI coding assistant talk directly to Alibaba Yunxiao — browse projects, track iterations, review code, and monitor pipelines without leaving your IDE.
Read-only by default, safety first. 170 of 182 tools are read-only queries. 12 write operations require explicit read_only=false.
What can you do?
| Scenario |
Tools |
| 📋 Project Management |
Browse projects, work items, iterations, milestones, members; create/update work items |
| 🔍 Code Review |
Browse repositories, branches, commits, merge requests, change requests; create/close/merge CRs |
| 🚀 Pipelines |
List pipelines, runs, build tasks; approve/reject validation gates |
| 📦 Release Management |
Browse applications, environments, releases, change orders, resources |
| 🧠 Knowledge Base |
Browse Lingma knowledge bases, members, files |
| 🤖 AI Adoption |
Analyze team Lingma usage |
Write operations (create/update work items, manage CR/MR, pipeline approvals) require read_only=false.
Quick Start
npx (zero install)
npx -y @futuretea/yunxiao-mcp-server
With token:
YUNXIAO_MCP_ACCESS_TOKEN=<your-token> npx -y @futuretea/yunxiao-mcp-server
Docker
Stdio mode (default):
docker run -i --rm -e YUNXIAO_MCP_ACCESS_TOKEN=<your-token> \
ghcr.io/futuretea/yunxiao-mcp-server:latest
HTTP mode:
docker run --rm -p 3000:3000 -e YUNXIAO_MCP_ACCESS_TOKEN=<your-token> \
ghcr.io/futuretea/yunxiao-mcp-server:latest --port 3000
Build from source
make build
YUNXIAO_MCP_ACCESS_TOKEN=<your-token> ./bin/yunxiao-mcp-server
IDE Setup
See MCP Client Config for Claude, Cursor, and other IDE setup examples.
| Domain |
Tools |
Access |
Description |
| Projex |
47 |
43 read-only + 4 write |
Projects, iterations, work items, milestones, test cases |
| Codeup |
37 |
31 read-only + 6 write |
Repositories, branches, commits, MR, CR, code review |
| Flow |
18 |
16 read-only + 2 write |
Pipelines, runs, build tasks, validation |
| Appstack |
62 |
58 read-only + 4 write |
Applications, environments, releases, change orders |
| Platform |
18 |
read-only |
Organizations, departments, members, roles |
| Packages |
3 |
read-only |
Artifact repositories and versions |
| Lingma |
6 |
read-only |
Knowledge bases and usage |
| API |
1 |
read-only |
Generic API fallback |
| Meta |
1 |
read-only |
Tool discovery |
Enhanced tools aggregate multiple API calls into single operations, reducing AI round-trips.
| Tool |
What it combines |
get_project_overview |
Project info + members + sprints + milestones + labels |
get_project_workitem_detail |
Work item + activities + comments + attachments + relations |
get_repository_overview |
Repository + default branch + recent commits + recent MRs |
get_change_request_overview |
CR detail + patch sets + comments |
get_pipeline_overview |
Pipeline info + latest run + history |
Full list: Enhanced Tools Index
Configuration
Priority: flags > environment variables > config file > defaults.
Required
| Variable |
Description |
YUNXIAO_MCP_ACCESS_TOKEN |
Yunxiao access token |
Optional
| Variable |
Description |
Default |
YUNXIAO_MCP_BASE_URL |
API base URL |
https://openapi-rdc.aliyuncs.com |
YUNXIAO_MCP_SSE_BASE_URL |
Public SSE base URL (reverse proxy) |
— |
YUNXIAO_MCP_INSECURE_SKIP_TLS_VERIFY |
Skip TLS verify (internal only) |
false |
Legacy aliases: YUNXIAO_ACCESS_TOKEN, YUNXIAO_API_BASE_URL.
| Flag |
Default |
Purpose |
--read-only |
true |
Set false to enable write tools |
--compact |
true |
Hide raw tools with enhanced alternatives (set false to show all) |
--enabled-tools |
— |
Explicit tool allow-list |
--disabled-tools |
— |
Explicit tool deny-list |
--enable-domains |
— |
Domain allow-list |
--disable-domains |
— |
Domain deny-list |
Config File
./bin/yunxiao-mcp-server --config config.example.yaml
Per-Request Token (HTTP/SSE)
curl -H "x-yunxiao-token: <token>" http://localhost:3000/mcp
# or
http://localhost:3000/sse?yunxiao_access_token=<token>
HTTP Endpoints
| Endpoint |
Purpose |
/mcp |
Streamable HTTP MCP |
/sse |
SSE MCP |
/message |
SSE message endpoint |
/healthz |
Health check (503 if no tools registered) |
Security
- Read-only by default: 170 tools safe for exploration without write access.
- Explicit write opt-in: 12 write tools require manual
read_only=false.
- Per-request token: HTTP/SSE support request-level token override for multi-tenant use.
- No sensitive endpoints: Admin audit logs, PAT queries, and other high-privilege endpoints are excluded.
Development
make fmt # gofmt
make tidy # go mod tidy
make lint # go vet + gofmt
make test # go test ./...
make build # build binary
make smoke # smoke test
make ci # full CI
Coverage threshold: 98%. Run make coverage-check.
Documentation
License
MIT