gate-info-risk
Overview
Risk-oriented primary skill for Gate.info + Gate.news, executed via gate-cli. Consolidates token-contract risk, address compliance risk, and project-level incident risk into a single five-section verdict.
Runtime requirements
- CLI:
gate-cliv0.5.2 onPATH. - Shell (optional): for
scripts/update-skill.*only. - Credentials: as required by
gate-cli/ preflight for the commands you invoke.
Core capabilities
| Capability | Typical trigger | Playbook id |
|---|---|---|
| Token contract risk (honeypot, tax, holder concentration) | "Is PEPE safe on eth" / "is this a honeypot" | token_risk |
| Address compliance risk (labels, OFAC, blacklist) | "Is this address safe" / "sanctioned / OFAC" | address_risk |
| Project-level incident risk (exploit, depeg, regulatory) | "any security or compliance incident on {project}" | project_risk |
Address risk limitation:gate-cli v0.5.2does NOT shipinfo compliance check-address-risk. Address risk is sourced frominfo onchain get-address-infolabels. When those are missing the verdict is always UNABLE_TO_ASSESS (scope limited), never LOW. See skills/gate-info-risk/references/troubleshooting.md.
Inputs / outputs
- Input: natural-language query. Required slots:
- token_risk → token (contract or ticker) + chain - address_risk → address + chain - project_risk → symbol
- Output: five-section structured report in the user’s locale; Section 1 states HIGH / MEDIUM / LOW / UNABLE_TO_ASSESS (may be localized in the user-facing report).
Routing (when NOT to use)
| Intent | Route to |
|---|---|
| "Give me a general analysis of {coin}" (no safety framing) | gate-info-research |
| "Trace this address / smart money behavior" (not risk-first) | gate-info-web3 |
| "Why did it drop / community view on the incident" | gate-news-intel |
Full cross-skill matrix: skills/_shared/routing.md.
Acceptance criteria
- Preflight contract followed (skills/_shared/preflight.md).
- Every
gate-clicommand appears ingate-cli info list/gate-cli news list.check-address-riskis NOT called. - Report has all five sections in the required order (product requirements spec).
- Missing data is labelled scope limited and the verdict degrades to UNABLE_TO_ASSESS — never upgrades to LOW without evidence.
- Verdict ordering inside Section 2 is strictly high → medium → low.
Source
- Skill spec: SKILL.md
- Playbook: playbooks/gate-info-risk.yaml
- References: references/scenarios.md, references/cli-reference.md, references/troubleshooting.md
- Repository:
gate-github-skills(layout may mirrorgate/gate-skills).