GEO-focused fact-checking and evidence collection assistant for written content. Use this skill whenever the user wants to verify factual claims (numbers, da...
Security Analysis
high confidenceThe skill's files, instructions, and runtime requirements are consistent with a fact‑checking assistant: it uses web search/fetch and a small local claim-extraction helper, and requests no extra credentials or installs.
The name and description match the included materials: SKILL.md describes GEO fact-checking workflows, references/ contains claim and pattern guides, evals/ contains example prompts, and scripts/claim_extractor.py provides a small, local helper. There are no unrelated credentials, binaries, or install requirements that would be out of place for a fact-checking skill.
Runtime instructions explicitly call for using web search and fetch tools and reading the user's draft and the bundled reference files. That is appropriate for the stated purpose. The skill also permits using the included scripts/claim_extractor.py to parse text; executing local helper scripts is expected for this use case but means code in the skill may run in the agent environment—the included script is a simple, non-obfuscated Python extractor with no network calls. Users should avoid placing sensitive secrets or PII in drafts that the agent will send to external web tools.
No install specification is provided (instruction-only with a local helper script). This minimizes disk-write and downloader risks. Nothing is fetched from external URLs by the skill package itself.
The skill does not request environment variables, credentials, or config paths. The web search/fetch actions the SKILL.md expects are standard for fact-checking and do not require extra secrets in the package.
always:false and user-invocable:true. The skill can be invoked autonomously by agents (disable-model-invocation:false), which is the platform default; it does not request permanent inclusion or modify other skills. If you prefer to avoid autonomous use, consider adjusting agent or skill invocation policies.
Guidance
This skill appears to do what it says: extract candidate factual claims and verify them using web search/fetch and the provided reference files. Before installing, consider: (1) The skill will use the agent's web tools to fetch external pages — ensure those tools and endpoints are trusted. (2) The skill may execute the included claim_extractor.py helper; that script is small and contains only local text parsing (no network or obfuscated code), but execution of any code should match your security policy. (3) Don’t include sensitive credentials or PII in drafts you ask the agent to check, because the skill explicitly reads user content and sends queries to external web fetch/search tools. (4) If you do not want the skill to be invoked autonomously, change agent/skill invocation settings. Overall the package is internally coherent and proportionate to its purpose.
Latest Release
v0.1.0
Initial release — introduces a GEO-focused fact-checking assistant for rigorous content verification and evidence collection. - Identifies and extracts significant factual claims for enhanced trust (numbers, dates, rankings, market data, quotes). - Guides the verification of claims using authoritative and up-to-date sources, with prioritization of accuracy and transparency. - Documents workflow: extract claims, plan checks, run fact checks, compare evidence, classify results, and propose robust corrections. - Explicitly flags outdated, unclear, or contradicted information, suggesting clear, evidence-backed improvements. - Optimized for use with AI-citable content: reports, comparison pages, and data-driven articles.
Popular Skills
Published by @geoly-geo on ClawHub
