truthcheck

Verify claims, fact-check content, check URL trustworthiness, and trace claims to their origin using the TruthCheck CLI. Use when: (1) user asks to fact-check or verify a claim, (2) user wants to check if a URL/source is trustworthy, (3) user wants to trace where a claim originated, (4) user asks about misinformation or content verification. Requires: pip install truthcheck

Safety Notice

This listing is from the official public ClawHub registry. Review SKILL.md and referenced scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "truthcheck" with this command: npx skills add baiyishr/truthcheck

TruthCheck Skill

AI content verification CLI. Verify claims, check URLs, and trace misinformation.

Commands

Verify a claim

truthcheck verify "claim text" --llm gemini
  • Returns TruthScore 0-100 with breakdown (publisher, content, corroboration, fact-checks)
  • --llm is optional but improves accuracy
  • Add --json for structured output

Check URL trustworthiness

truthcheck check https://example.com
truthcheck check "text with multiple URLs"
truthcheck check -f file.txt
  • Detects hallucinated URLs that don't exist
  • Rates publisher credibility

Trace a claim to its origin

truthcheck trace "claim text"           # balanced, ~1-2 min
truthcheck trace "claim text" --quick   # fast, ~30 sec
truthcheck trace "claim text" --deep    # thorough, ~3-5 min
  • Finds original source and builds propagation tree

Look up a publisher

truthcheck lookup "publisher name"

Sync publisher database

truthcheck sync

Run periodically to keep publisher ratings current.

Interpreting TruthScore

ScoreLabelMeaning
80-100TRUEStrong evidence supports the claim
60-79POSSIBLY TRUESome supporting evidence, inconclusive
40-59UNCERTAINMixed or insufficient evidence
20-39POSSIBLY FALSEEvidence leans against the claim
0-19FALSEStrong evidence contradicts the claim

LLM Integration (Optional)

TruthCheck works without any LLM, but adding --llm improves content analysis accuracy.

truthcheck verify "some claim" --llm gemini    # recommended, fast & free tier
truthcheck verify "some claim" --llm openai    # GPT models
truthcheck verify "some claim" --llm anthropic # Claude models
truthcheck verify "some claim" --llm ollama    # local models, fully offline

Privacy: API keys are stored in your local environment only. TruthCheck never sends your keys to any external service — they are used solely for direct API calls to the respective LLM provider.

Environment Variables

VariableWhen neededDescription
GEMINI_API_KEY--llm geminiGoogle AI API key (free tier available)
OPENAI_API_KEY--llm openaiOpenAI API key
ANTHROPIC_API_KEY--llm anthropicAnthropic API key
BRAVE_API_KEY--search braveBrave Search API key

No keys needed for --llm ollama (runs locally) or default DuckDuckGo search.

Tips

  • Verify commands can take 15-60 seconds depending on search results
  • Without --llm: basic scoring using publisher reputation, corroboration, and fact-checks
  • With --llm: adds AI content analysis for better accuracy
  • --search brave gives better search results than default DuckDuckGo
  • For batch verification, loop through claims individually

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

Coding

Agent Dev Workflow

Orchestrate coding agents (Claude Code, Codex, etc.) to implement coding tasks through a structured workflow. Use when the user gives a coding requirement, f...

Registry SourceRecently Updated
Coding

Cortex Engine

Persistent cognitive memory for AI agents — query, record, review, and consolidate knowledge across sessions with spreading activation, FSRS scheduling, and...

Registry SourceRecently Updated
Coding

Skill Blocker - 安全守卫

Blocks execution of dangerous commands and risky operations like destructive deletions, credential theft, code injection, and unauthorized system changes to...

Registry SourceRecently Updated
014
Profile unavailable