Matic Trades
This skill tells you how to call Matic Trades from OpenClaw using the bundled script. Keep instructions short: run the script, pass the user’s words as --prompt, return a clear summary (and image URLs when present).
How OpenClaw uses this
- Skills are markdown + optional files; the model follows this file when the task matches the description.
- Prefer
python3 "{baseDir}/scripts/matic.py" …so requests stay consistent with the real API (no hand-rolled JSON in chat). - After the script prints JSON, summarize for the user; don’t dump raw JSON unless they ask.
Docs: OpenClaw — Creating skills, Skills loading.
Config
| Variable | Required | Purpose |
|---|---|---|
MATIC_API_KEY |
Yes | Bearer token (same as dashboard API key). Alias: MATIC_TRADES_API_KEY. |
MATIC_TRADES_API_BASE |
No | Default https://api.matictrades.com/api/v1 (no trailing slash). Override if your deployment uses another host. |
Never print or log the API key.
1) Toolbox — AI_PICK (default for “use Matic tools”)
What the API does: POST /toolbox/execute with mode: AI_PICK. The server uses the prompt alone to decide which tools to run (e.g. chart_analysis, market_news, social_search), then executes them. context is not required and is not what drives tool selection in normal use—put everything important in the prompt (tickers, timeframe, what they want).
Minimal request body (this is correct):
{ "mode": "AI_PICK", "prompt": "User’s full request in natural language" }
Run:
python3 "{baseDir}/scripts/matic.py" toolbox --prompt "USER_TEXT_HERE"
Examples (prompt-only):
toolbox --prompt "Unusual options activity and any relevant news today"toolbox --prompt "Chart and sentiment for NVDA — focus on near-term levels"toolbox --prompt "Crypto whale moves and what people are saying on social"
Do not require or invent a context object for AI_PICK unless you are implementing a documented advanced API feature (e.g. custom_handlers from Matic’s API docs). For ClawHub users, prompt-only is the supported path.
2) Data — SMART_SEARCH (Twelve Data via natural language)
What the API does: POST /data/execute with mode: SMART_SEARCH. The server maps the prompt to an allowed Twelve Data endpoint and parameters.
Run:
python3 "{baseDir}/scripts/matic.py" data --prompt "USER_TEXT_HERE"
Optional hints only if the user gave them: --symbol AAPL, --interval 1day. The API still centers on prompt for SMART_SEARCH.
3) Agent charting — autonomous charts + analysis
What the API does: POST /agent/chart/autonomous — multi-step chart agent; can return image URLs or base64 in snapshots.
Run:
python3 "{baseDir}/scripts/matic.py" chart --prompt "USER_TEXT_HERE" --images url
Use --images url when the client can open links; b64 or both if the UI supports inline images.
Optional: --max-actions N, --model MODEL (see references/api.md).
Which command when?
| User goal | Command |
|---|---|
| Broad research, multiple possible tools, news/social/chart mix | toolbox (AI_PICK, prompt only) |
| “Get RSI / price / indicator / series for …” in plain English | data (SMART_SEARCH) |
| “Walk through charts”, “show me the chart”, visual narrative | chart |
Chaining: Only if it clearly helps—e.g. data for numbers, then toolbox for synthesis, or chart for visuals. Avoid redundant double calls.
Response handling
- Parse JSON stdout; explain tool_runs / analysis / data in plain language.
- If
snapshots.pre_url,snapshots.post_url, or URLs appear in outputs, give the user those links.
Errors
- 401 — Missing/wrong
MATIC_API_KEY. - 429 — Rate limit; slow down or check plan.
- Quota / billing messages — Point to Matic dashboard.
Files
references/api.md— request/response shapes.README.md— install, ClawHub upload, env setup.
Scope
Only call matic.py with user-supplied prompt text (sanitized for shell if needed). HTTP targets only MATIC_TRADES_API_BASE paths used by this script. No arbitrary curl to other hosts with user data.