What does this skill do, and when should you use it?
This is a read-only data skill defined at skills/polymarket/SKILL.md in the HKUDS/AI-Trader repository, for querying public information on Polymarket prediction markets. It teaches an agent to resolve a market via the Gamma API using a slug, conditionId, or token_id, pair outcome tokens with clobTokenIds to identify a specific outcome, then fetch best bid/ask from the CLOB orderbook API and derive a mid price. The skill performs no trading itself; AI-Trader is used only to publish simulated trades and share signals. The repository is MIT-licensed per its README badge, and Polymarket paper trading went live on 2026-03-03.
Calls the Gamma markets API (https://gamma-api.polymarket.com/markets) via curl to resolve markets by slug or conditionId; reads question, slug, outcomes, and clobTokenIds fields and pairs each outcome with its token ID; calls the CLOB orderbook API (https://clob.polymarket.com/book) by token_id to read best bid/ask and derive a mid price; publishes a simulated trade to AI-Trader in the specified JSON shape (market, symbol, outcome, token_id, price, quantity, etc.).
- An agent for a prediction-market analyst that needs live public orderbook prices for a Polymarket market
- Developers building their own Polymarket data pipeline who want discovery traffic going straight to official Polymarket APIs
- AI-Trader users who want to confirm a market and its specific outcome token (e.g. the token_id for Yes/No) locally before publishing a simulated trade
- Researchers comparing mid prices across multiple markets who need to batch-resolve conditionIds to market metadata
- Community agents syncing Polymarket trade signals to the AI-Trader platform for discussions and leaderboards
How do you install this skill?
- This assessment is a purely static source review; no API calls were executed and example outputs are independently unverified.
- Core functionality depends entirely on Polymarket's gamma-api and CLOB endpoints, which may be unreachable from mainland-China networks; verify reachability before use.
- Publishing trades to the AI-Trader platform requires registration and exposes an agent identity on an external service — an external effect users should consciously accept.
- The skill file has no version number or changelog; Polymarket API field changes could invalidate instructions, so verify actual responses at runtime.
- The example token_id and price are placeholders; copying them verbatim will not produce valid results.
- Shell / CLI
- Network access
curl or equivalent HTTP client
The skill lives at skills/polymarket/SKILL.md in the AI-Trader repository. Clone the repo, or place the skills/polymarket folder into your skills directory following the Agent Skills standard (the exact install path is not documented in the source). For the broader platform, the README says an agent can register by reading https://ai4trade.ai/SKILL.md.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/HKUDS/AI-Trader.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/polymarket" ~/.claude/skills/
rm -rf "$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?
Have the agent run the skill's curl commands: first GET https://gamma-api.polymarket.com/markets?slug=<slug> or ?conditionId=<id> to resolve the market, pairing outcomes with clobTokenIds; then GET https://clob.polymarket.com/book?token_id=<id> for best bid/ask and derive the mid; finally publish a simulated trade to AI-Trader using the skill's JSON structure (market: "polymarket", symbol, outcome, token_id, etc.).
What are this skill's strengths and limitations?
- Market-discovery traffic goes directly to Polymarket's official public APIs with no third-party relay
- Purely read-only queries — no Polymarket account or private keys needed
- Clear, copyable curl commands and explicit field-mapping rules (outcomes[i] paired with clobTokenIds[i])
- Hosted in the MIT-licensed HKUDS/AI-Trader repo, which is actively maintained
- Covers only read-only market data and simulated-trade publishing; it places no real orders
- No test suite or error-handling guidance in the skill itself
- Depends on Polymarket public API availability and rate limits, which the source does not document
- Repository metadata lists license as unknown; only the README badge indicates MIT
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 |
|---|---|---|---|---|
| Polymarket Public Data Skill this page | 50 · Use with care | ★ 23k | 4mo ago | — |
| Market Intel — AI-Trader Read-Only Market Data Skill | 47 · Use with care | ★ 23k | 4mo ago | — |
| Lighter Agent Kit | 57 · Use with care | ★ 67 | 2mo ago | MIT |
| AI-Trader Signal Skill | 40 · Not recommended | ★ 23k | 4mo ago | — |
| AI-Trader Copy Trading Skill | 40 · Not recommended | ★ 23k | 4mo ago | — |
How did FollowSkills review this skill?
The skill itself performs read-only curl calls to public Polymarket APIs — no credentials, no write permissions — with transparent data-flow disclosure and explicit routing policy; deductions: publishing trades to the AI-Trader platform involves registration and external service dependency, with no user-confirmation mechanism, no rollback guidance, and unverified publisher identity.
Instructions are self-consistent, curl examples are well-formed, and the resolve path (slug/conditionId → token_id → orderbook) is clear; deductions: static review without execution, API field/response structures unverified, no error handling or failure-feedback guidance for API changes, rate limits, or empty responses.
Trigger scenarios are clear (when Polymarket metadata or public prices are needed), with clear inputs, outputs and a recommended agent flow; deductions: no declared capability boundaries or non-fit ranges, core function depends entirely on overseas Polymarket APIs likely unreachable from mainland-China networks, and the doc discloses neither this limitation nor Chinese-language support.
Well-structured with good progressive disclosure (usage → endpoints → examples → flow) and stable naming; deductions: no version number, changelog, or known-limitation disclosure in the skill file, no stated maintenance ownership or update path; repo README shows an MIT badge but the skill file carries no license info, and the publisher is unverified.
Claims direct market resolution and price reading via public APIs while avoiding AI-Trader traffic routing — a clear marginal benefit; deductions: static review cannot confirm output direct usability, example token_id ('123456789') is a placeholder, example price is 0, and evidence of real-world usability is thin.
Primary material (skill doc and repo source) is auditable and endpoints are publicly checkable; deductions: static read only, no execution evidence, no dedicated test coverage for this skill's key paths (existing tests cover unrelated server features), and no independent reproduction or cross-source corroboration.
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 →