agentleFS
Baba01hacker666Baba01hacker666/dummyskills/skills/claim-checker/SKILL.md

claim-checker

Skill0 starsChanged 2 months agoMITView on GitHub

Audits specific factual claims against provided reference documents, labeling each as supported, refuted, or unverified.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 4 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when:

    line 14
  2. Very specific:

    Do not use to write a general summary.

    line 20
  3. Very specific:

    Do not use to compare competing options.

    line 21
  4. Very specific:

    If claims is missing: stop and ask: "Please provide the list of claims to verify."

    line 28
  5. Very specific:

    If reference_documents is missing: stop and ask: "Please provide the reference text to check claims against."

    line 29
  6. Very specific:

    Do not use external world memory to verify claims if reference documents are silent.

    line 30
  7. Distinctive:

    Follow these numbered actions:

    line 33
  8. Distinctive:

    If a claim is partially true and partially false: split into Claim A (Supported) and Claim B (Refuted).

    line 46
  9. Very specific:

    Never label a claim SUPPORTED without an exact quotation from reference_documents.

    line 47
  10. Distinctive:

    Stop and ask the user when:

    line 77

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

Claim Checker

Audits explicit factual assertions against provided source documents. Categorizes each statement as SUPPORTED, REFUTED, or UNVERIFIED with exact quotation evidence.

1. Purpose

Prevent misinformation and confirm claims before publication or integration. A "Claim" is an assertion of fact. "Evidence" is an exact matching excerpt from reference text.

2. When to Use

Use this skill when:

  • Reviewing marketing copy, documentation, or reports against technical specs.
  • User asks: "Check if these statements are true according to the docs."
  • Pre-release verification of release announcements.

3. When Not to Use

  • Do not use to write a general summary; use factual-summarizer.
  • Do not use to compare competing options; use option-comparator.

4. Required Inputs

  • claims: List of 1 to 10 specific statements to audit.
  • reference_documents: Authoritative text or document excerpts provided as ground truth.

5. Missing-Input Behavior

  1. If claims is missing: stop and ask: "Please provide the list of claims to verify."
  2. If reference_documents is missing: stop and ask: "Please provide the reference text to check claims against."
  3. Do not use external world memory to verify claims if reference documents are silent; mark as UNVERIFIED (Not in provided source).

6. Procedure

Follow these numbered actions:

  1. Action 1: Number and Isolate Each Claim
    • Split compound claims into single verifiable units.
  2. Action 2: Search Reference Text for Relevant Evidence
    • Extract exact quote matching the subject matter.
  3. Action 3: Assign Verification Label
    • SUPPORTED: Direct quotation confirms the claim.
    • REFUTED: Direct quotation directly contradicts the claim.
    • UNVERIFIED: Reference text contains no relevant evidence.
  4. Action 4: Emit Verification Table
    • Format per Section 8.

7. Decision Rules

  • If a claim is partially true and partially false: split into Claim A (Supported) and Claim B (Refuted).
  • Never label a claim SUPPORTED without an exact quotation from reference_documents.

8. Output Format

Emit the audit using this structured template:

### Claim Verification Audit

| # | Claim | Status | Quoted Evidence from Source |
| :- | :--- | :--- | :--- |
| 1 | [Claim statement] | **SUPPORTED** | "[Exact quote from source]" |
| 2 | [Claim statement] | **REFUTED** | "[Exact quote showing contradiction]" |
| 3 | [Claim statement] | **UNVERIFIED** | *No mention in provided reference text.* |

#### Audit Summary
- **Total Claims**: [Count]
- **Supported**: [Count] | **Refuted**: [Count] | **Unverified**: [Count]

9. Verification Checklist

  • Every SUPPORTED or REFUTED claim has an exact quote.
  • Claims not in the source are marked UNVERIFIED.
  • No assertions are judged based on unstated assumptions.

10. Common Failures and Recovery

Failure SymptomCauseRecovery Action
Assuming claim is trueWorld knowledge biasForce check: is there a quote in the provided text? If no, label UNVERIFIED.
Paraphrasing quotesInexact citationsUse verbatim quotation enclosed in quotation marks.

11. Stop or Escalation Conditions

Stop and ask the user when:

  • Reference documents are encrypted or empty.

12. Examples

Example 1: Successful Execution

Input:

Claims:
1. The server supports TLS 1.3.
2. Default port is 8080.
3. Redis is required.
Reference: "The gateway binds to port 8080 by default. It mandates TLS 1.3 for all incoming connections. An optional in-memory caching engine is used; no external cache server is required."

Output:

### Claim Verification Audit

| # | Claim | Status | Quoted Evidence from Source |
| :- | :--- | :--- | :--- |
| 1 | The server supports TLS 1.3 | **SUPPORTED** | "It mandates TLS 1.3 for all incoming connections." |
| 2 | Default port is 8080 | **SUPPORTED** | "The gateway binds to port 8080 by default." |
| 3 | Redis is required | **REFUTED** | "no external cache server is required." |

#### Audit Summary
- **Total Claims**: 3
- **Supported**: 2 | **Refuted**: 1 | **Unverified**: 0

Example 2: Ambiguous or Missing Input

Input:

Claims: Python was released in 1991.
Reference: (Empty text)

Output:

### Claim Verification Audit

| # | Claim | Status | Quoted Evidence from Source |
| :- | :--- | :--- | :--- |
| 1 | Python was released in 1991 | **UNVERIFIED** | *Reference document was empty; no source evidence provided.* |

#### Audit Summary
- **Total Claims**: 1
- **Supported**: 0 | **Refuted**: 0 | **Unverified**: 1

Example 3: Expected Final Output

Output:

### Claim Verification Audit

| # | Claim | Status | Quoted Evidence from Source |
| :- | :--- | :--- | :--- |
| 1 | Maximum file upload is 25 MB | **SUPPORTED** | "Payloads exceeding 25 MB will return status 413." |
| 2 | Webhooks retry up to 10 times | **UNVERIFIED** | *No retry parameters mentioned in provided configuration section.* |

#### Audit Summary
- **Total Claims**: 2
- **Supported**: 1 | **Refuted**: 0 | **Unverified**: 1

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.