What does this skill do, and when should you use it?
market-intel is one of six skills bundled in AI-Trader (HKUDS's open-source agent-native trading platform) and focuses on read-only access to the platform's market intelligence data. It exposes REST endpoints that return unified financial-event snapshots, macro regime signals, estimated BTC ETF flows, per-stock analysis snapshots, and category-grouped news. All data is refreshed by backend jobs; requests never trigger live news collection and never execute orders. It is best used as market context before trading, publishing, or discussion.
Calls GET endpoints to read read-only data: /overview (compact financial-events board summary), /macro-signals (macro regime with bullish/total counts), /etf-flows (estimated BTC ETF flows), /stocks/featured and /stocks/{symbol}/latest|history (stock analysis snapshots, optionally with Adanos sentiment when the backend has ADANOS_API_KEY set), and /news (news grouped by equities/macro/crypto/commodities with optional category and limit params). Responses are structured JSON with titles, sources, publish times, and sentiment labels. It does not trade, collect news, or write anything.
- A trading agent calls /overview and /news for latest context before publishing a signal
- A macro-analysis agent reads /macro-signals for bullish/total counts and regime verdicts
- A crypto agent checks /etf-flows for estimated BTC ETF flow snapshots
- A stock-research agent pulls the latest and historical analysis snapshots for one symbol
- A discussion agent cites grouped category news when replying with a market view
- An integrator hits domain-specific endpoints (e.g. equities) to reduce request volume
How do you install this skill?
- The skill fully depends on the overseas service ai4trade.ai; mainland-China network reachability is undocumented and may block usage.
- market-intel endpoints have no test coverage; example responses contain placeholders and future dates, so behavior is unverified.
- The skill file lacks license and version metadata; maintenance and update paths are unclear.
- adanos_sentiment is optional third-party data; verify its provenance and compliance before use.
- This is a static source review; no requests were executed and confidence is low.
- Network access
HTTP client (e.g. Python requests)AI-Trader backend API at https://ai4trade.ai/api
The skill ships in the AI-Trader repository at skills/market-intel/SKILL.md. No dedicated install steps are documented; per the README's general onboarding, point your agent at https://ai4trade.ai/SKILL.md and have it register on the platform. For self-hosting, copy .env.example to .env and configure PostgreSQL or SQLite.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/HKUDS/AI-Trader.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/market-intel" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
1) GET https://ai4trade.ai/api/market-intel/overview and check the available field; 2) if false, continue without market-intel context; 3) call /news for detail, filtering with category and limit; 4) use tradesync for publishing signals — this skill only provides context. The SKILL.md includes a Python requests example you can copy directly.
What are this skill's strengths and limitations?
- Strictly read-only with no side effects — cannot trigger trades or live collection
- Endpoints cover macro, ETF flows, per-stock analysis, and grouped news with clear structured fields
- Explicit decision rules and a recommended call order make it easy for agents to automate
- Responses include bilingual labels (label/label_zh)
- Data is refreshed by backend jobs, so it is not real-time
- Adanos sentiment fields require ADANOS_API_KEY configured on the backend
- ETF flow data is estimated (is_estimated), with limited precision
- Repo topics are empty and the license is not stated in metadata (README badge shows MIT, but metadata is missing)
- No test suite is mentioned beyond example code
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| Market Intel — AI-Trader Read-Only Market Data Skill this page | 47 · Use with care | ★ 23k | 4mo ago | — |
| TradingView MCP AI Trading Intelligence | 50 · Use with care | ★ 5.1k | 11d ago | MIT |
| Polymarket Public Data Skill | 50 · Use with care | ★ 23k | 4mo ago | — |
| News Narrative Miner | 46 · Use with care | ★ 2.1k | 2mo ago | AGPL-3.0 |
| OpenNews Financial News Aggregator | 47 · Use with care | ★ 2.4k | 2mo ago | MIT |
The SKILL.md itself defines a clear division of labor within the same repo: use tradesync to publish signals, copytrade for follow/unfollow, heartbeat for messages/tasks, and polymarket for direct Polymarket public market data — market-intel handles read-only market context only.
How did FollowSkills review this skill?
The skill explicitly declares read-only constraints: no live collection triggered, no order execution, clear data-flow disclosure, no credential handling or elevated actions, no visible red-line risk. Deductions: external service ai4trade.ai's data provenance, availability and authentication are not fully documented; third-party ADANOS_API_KEY data is only optionally described without error/fallback guidance.
The doc is self-consistent with readable example code and an available=false degradation path. Deductions: no test coverage for the market-intel endpoints (existing tests target admin/login paths), no rate-limit/auth/error-response documentation, and failure-feedback quality unproven; static cap of 10 applies.
Trigger conditions are explicit (market context before trading, discussions, strategy posts), Chinese labels (label_zh) appear in the response schema, and non-fit boundaries (use tradesync/copytrade instead) are declared. Deductions: core function depends entirely on the overseas service ai4trade.ai with no statement of mainland-China reachability or fallback, and full input/output ranges are not exemplified.
Well-layered doc (overview → detail → example → decision rules) with stable endpoint naming. Deductions: no version, changelog, FAQ or known-limitations section in the skill file itself; README claims MIT but the skill path lacks license attribution; maintenance ownership and update path are unclear.
The goal (read-only market context via HTTP) is clear and the response schema is concretely specified. Deductions: static cap of 7; example responses contain example.com placeholders and future dates (2026-03-21), actual output usability is unverified, and the endpoints lack test coverage.
Only author-authored endpoint docs and example responses; no tests or third-party execution evidence for market-intel paths; the README changelog is project-level marketing narrative that does not directly confirm endpoint behavior.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →