
Fact-Check Claims Skill
Verifies every factual claim in content against live sources before it ships.
- Repository
- garrytan/gbrain
- Stars
- 31,447
- Licence
- MIT
- Files
- 1 file
What it does
When installed, the agent goes through a draft claim by claim, pulling out names, numbers, quotes, dates and superlatives, then checks each one against live web sources rather than its own memory. It labels each claim on a six-level confidence scale, suggests corrections or hedged wording, and produces a pass/fail report. For outputs generated from a brain or database, it also re-derives key claims through a separate query path before allowing delivery.
Great for
- Writers and editors checking an essay, article or blog post before publishing
- Anyone producing a briefing or report that must be factually reliable
- Teams wanting a repeatable checklist for verifying AI-generated output
Skip it if
- Opinion pieces or creative fiction, where the skill deliberately skips opinions and predictions and adds little value
- Situations needing guaranteed truth: it depends on the quality and availability of web sources and can still miss errors or report false negatives
- Highly confidential content, since claims and some context may be sent to web search and fetch tools
Try asking
- “Fact check this draft article before I publish it”
- “Verify the facts and statistics in this report claim by claim”
- “Is this accurate? Check every number and quote in this press release”
- “Re-derive every claim in this summary generated from our database”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the MIT licence and an attribution note.
fact-check.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip fact-check.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip fact-check.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Provide the full text and any sources you already have so the agent can check them first.
- Expect uncertain claims to be hedged rather than confirmed; review any UNVERIFIED or DISPUTED items yourself.
- Do not rely on the skill alone for legal or defamation review; have a human editor check high-stakes claims.
What's inside
- SKILL.md
Redistributed unmodified from garrytan/gbrain under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.