Uno MCP Tools

Call MCPMarket's REST API directly via curl to search and invoke 2000+ tools. No packages to install.

Prerequisites

• curl (pre-installed on most systems)

Authentication

# 1. Request a device code
curl -s -X POST https://mcpmarket.cn/oauth/device/code \
  -d "client_id=skill-agent&scope=mcp:*"

You MUST display the exact verification_uri and user_code from the response to the user. Never construct or modify the URL yourself.

# 2. Poll for token after user authorizes (retry every 5s until access_token is returned)
curl -s -X POST https://mcpmarket.cn/oauth/token \
  -H "Content-Type: application/x-www-form-urlencoded" \
  -d "grant_type=urn:ietf:params:oauth:grant-type:device_code&device_code=DEVICE_CODE&client_id=skill-agent"

# 3. Store the token
mkdir -p ~/.uno && chmod 700 ~/.uno
echo "ACCESS_TOKEN_VALUE" > ~/.uno/token && chmod 600 ~/.uno/token

Verify login:

curl -s https://mcpmarket.cn/api/uno/verify-token \
  -H "Authorization: Bearer $(cat ~/.uno/token)"

Two-Step Invocation (Core Flow)

# Step 1: Search tools, get tool_name and inputSchema
curl -s "https://mcpmarket.cn/api/uno/search-tools?q=weather&mode=hybrid&limit=5" \
  -H "Authorization: Bearer $(cat ~/.uno/token)"

# Step 2: Call the tool
curl -s -X POST https://mcpmarket.cn/api/uno/call-tool \
  -H "Authorization: Bearer $(cat ~/.uno/token)" \
  -H "Content-Type: application/json" \
  -d '{"tool_name":"tonghu-weather.weatherArea","arguments":{"area":"Beijing"}}'

API Endpoints

All endpoints use Base URL https://mcpmarket.cn and require Authorization: Bearer <token>.

search-tools Parameters

Returns: tools[] (with tool, desc, inputSchema + stats fields), servers[]

tools[] stats fields (for selection guidance):

Search keyword tips: The backend vector index is built from tool function descriptions, so keywords should match what the tool does, not your query intent.

• ✅ weather — matches weather tool descriptions
• ❌ weather in Shanghai tomorrow — reads like a query, less likely to match tools

call-tool Parameters

Response structure:

{"content": [{"type": "text", "text": "<JSON string>"}], "isError": false}

content[0].text is itself a JSON string and needs to be parsed again. On error, isError is true and error contains the message.

Downstream OAuth: Some services (e.g. GitHub, Notion) require authorization on first call:

{"auth_required": true, "auth_url": "https://...", "state_id": "..."}

Open auth_url, complete authorization, then call again directly — the platform links the token server-side automatically.

categories (Browse by Category)

curl -s https://mcpmarket.cn/api/uno/categories \
  -H "Authorization: Bearer $(cat ~/.uno/token)"

Returns category names, counts, and representative server names. Useful when you're not sure which tool to use.

rate-server (Rate After Use)

After calling a tool or fetching a skill, submit a rating (0–5) to help improve search ranking:

# Rate a tool (recommended — tool-level granularity)
curl -s -X POST https://mcpmarket.cn/api/uno/rate-server \
  -H "Authorization: Bearer $(cat ~/.uno/token)" \
  -H "Content-Type: application/json" \
  -d '{"tool_name":"amap-maps.maps_weather","rating":4.5,"user_id":"YOUR_USER_ID"}'

# Rate a skill
curl -s -X POST https://mcpmarket.cn/api/uno/rate-server \
  -H "Authorization: Bearer $(cat ~/.uno/token)" \
  -H "Content-Type: application/json" \
  -d '{"skill_id":"abc123","rating":4.0,"user_id":"YOUR_USER_ID"}'

When call-tool or skills-fetch succeeds, a rating_hint is included in the response. Please rate proactively when you receive it.

credits (Balance + Recharge)

curl -s https://mcpmarket.cn/api/uno/credits \
  -H "Authorization: Bearer $(cat ~/.uno/token)"

Returns:

{
  "credits": 86,
  "daily_free_credits": 100,
  "daily_free_max": 100,
  "total_consumed": 14,
  "recharge_url": "https://mcpmarket.cn/billing"
}

When credits run low, show recharge_url to the user.

skills-search (Search Agent Skills)

curl -s -X POST https://mcpmarket.cn/api/uno/skills-search \
  -H "Authorization: Bearer $(cat ~/.uno/token)" \
  -H "Content-Type: application/json" \
  -d '{"q":"wechat article","mode":"hybrid","limit":5}'

Returns skills[], each containing:

skills-fetch (Fetch Full Skill Content)

curl -s -X POST https://mcpmarket.cn/api/uno/skills-fetch \
  -H "Authorization: Bearer $(cat ~/.uno/token)" \
  -H "Content-Type: application/json" \
  -d '{"skill_ids":["abc123","def456"]}'

Returns for each skill:

• skill_md: Full SKILL.md content
• files: File list
• download_url: ZIP download link
• repo_url: GitHub repository link

Maximum 10 skills per request.

⚠️ Parameter Construction Rules (Must Read)

Always read inputSchema before calling — never guess parameters.

Standard two-step flow (step 1 is mandatory):

# Step 1: search-tools, read inputSchema (always do this first)
curl -s "https://mcpmarket.cn/api/uno/search-tools?q=<keyword>&mode=hybrid&limit=5" \
  -H "Authorization: Bearer $(cat ~/.uno/token)"
# → Check required / minLength / field names / enum carefully

# Step 2: construct correct arguments from schema and call
curl -s -X POST https://mcpmarket.cn/api/uno/call-tool \
  -H "Authorization: Bearer $(cat ~/.uno/token)" \
  -H "Content-Type: application/json" \
  -d '{"tool_name":"<tool>","arguments":{<fill from schema>}}'

Workflow Tips

1. Search tools first, read inputSchema before calling — the most important step, never skip
2. Keywords match tool function, not query intent — weather ✅, tomorrow's weather in Shanghai ❌
3. Copy field names from schema — don't rely on memory; filters ≠ filter
4. Rate after successful calls — when rating_hint is returned, call rate-server to help improve rankings
5. Need a capability or guide — use skills-search to find, skills-fetch to load full content
6. No results — try English keywords, or switch to mode=semantic
7. Low credits — use credits to check balance, direct user to recharge_url

Credential Reference