Fact Check Before Trust

Fact-Check Before Trust

Completion verification checks that a task was done. This skill checks that the facts are correct. A confidently stated wrong number passes completion checks — this skill catches it.

When to Invoke

Before treating agent output as authoritative when it contains:

• Numbers or money — prices, quantities, measurements, statistics
• Dates and deadlines — filing deadlines, release dates, expiry dates
• Named entities — people, organisations, laws, product names
• Causal claims — "X causes Y", "because of Z"
• Superlatives — "the largest", "the only", "the most recent"

Skip for: code output, file operations, clearly self-contained tasks.

Verification Protocol

Step 1: Extract Claims

List every verifiable claim explicitly:

Claim 1: Widget costs $49/month
Claim 2: Free tier supports up to 1000 requests
Claim 3: API launched in March 2024

Step 2: Verify Each Claim

For each claim, check against a primary source:

• Official documentation or website
• Government/regulatory sources for legal claims
• Direct API response for technical claims

Step 3: Flag Discrepancies

For any claim that can't be verified or is wrong:

• State what the source says vs. what was claimed
• Correct the output before delivering
• Note confidence level: verified / unverified / corrected

Key Principle

Confidence is not evidence. An agent stating something firmly does not make it true. The more specific and consequential a claim, the more it needs verification.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: gb3h
• Source: gb3h/agent-skills
• License: MIT

Install and usage instructions live in the source repository linked above.