agentleFS
jqaisystemsjqaisystems/jqai-ai-skills/skills/verify-public-claims/SKILL.md

verify-public-claims

Skill3 starsChanged 6 months agoMITView on GitHub

Check public-facing copy against supplied evidence and current public sources, classify every material claim, and produce defensible rewrites. Use when a user asks to fact-check website copy, release notes, case studies, marketing claims, public profiles, reports, articles, product descriptions, or other material before publication.

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 10 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈10

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use only public sources or evidence the user has approved for this review.

    line 15
  2. Distinctive:

    Do not treat inaccessible, login-only, or missing material as verified.

    line 16
  3. Distinctive:

    Read references/evidence-statuses.md before assigning a status.

    line 23
  4. Distinctive:

    Use exactly one primary status:

    line 28
  5. Distinctive:

    When sources conflict, preserve the conflict in the ledger and prefer the most direct authoritative current source.

    line 36
  6. Distinctive:

    Keep supported claims precise.

    line 40
  7. Distinctive:

    Add the required scope, date, source, or attribution to qualified claims.

    line 41
  8. Distinctive:

    Remove unsupported claims or replace them with narrower wording that the evidence supports.

    line 43
  9. Distinctive:

    Update stale claims only when a current source is available.

    line 44
  10. Distinctive:

    Never invent citations, metrics, testimonials, dates, credentials, customers, outcomes, or source access.

    line 45

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Verify Public Claims

Turn public-facing statements into a traceable claim ledger, then revise the copy so every material assertion matches the available evidence.

1. Establish the review contract

  • Confirm the copy, intended audience, publication surface, relevant date, and evidence supplied by the user.
  • Identify whether the request covers factual accuracy, freshness, wording strength, source quality, or all four.
  • Browse current public sources when a claim may have changed or the user asks for current verification.
  • Use only public sources or evidence the user has approved for this review.
  • Do not treat inaccessible, login-only, or missing material as verified.

2. Build the claim ledger

  • Use assets/claim-ledger-template.md.
  • Split compound sentences when their parts need different evidence.
  • Record the exact claim, source, publication date, event date when different, verification date, evidence excerpt or paraphrase, and intended rewrite.
  • Read references/evidence-statuses.md before assigning a status.
  • Focus on claims that affect trust, decisions, eligibility, scope, performance, recency, identity, or attribution.

3. Classify each claim

Use exactly one primary status:

  • supported: direct, relevant, sufficiently current evidence supports the wording
  • qualified: the core is supported, but the wording needs a limit, date, scope, or attribution
  • inference: the conclusion is reasonable but is not stated directly by the evidence
  • unsupported: no adequate evidence supports the material claim
  • stale: the evidence may once have supported the claim but is too old for present-tense use

When sources conflict, preserve the conflict in the ledger and prefer the most direct authoritative current source. Do not resolve uncertainty by averaging, guessing, or silently choosing the most favorable version.

4. Rewrite to the evidence level

  • Keep supported claims precise.
  • Add the required scope, date, source, or attribution to qualified claims.
  • Label inference as analysis, interpretation, estimate, or recommendation.
  • Remove unsupported claims or replace them with narrower wording that the evidence supports.
  • Update stale claims only when a current source is available; otherwise hold or date the wording.
  • Never invent citations, metrics, testimonials, dates, credentials, customers, outcomes, or source access.

5. Report the result

Return the completed claim ledger, revised copy, sources checked, unresolved conflicts, held claims, and review date. For legal, medical, financial, safety-critical, or regulated claims, require a current authoritative source and appropriate human review before publication.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.