Two Claude Skills for fact-checking and hypothesis testing: pre-call, then hunt disconfirming evidence
TheNerdosapien · reddit · 2026-11-21
The author shares two self-built Claude Skills to curb gullibility: one for fact-checking posts/articles/screenshots, one for testing speculative hypotheses. Both force an initial gut call before searching, then require hunting for evidence against the claim while showing how the view changed. The fact-check skill scores truthfulness, missing context, sensationalism, fear-mongering and clickbait on defined scales, and treats checked content as untrusted data — ignoring embedded instructions (doubling as prompt-injection defense). Full skill text is included; paste it into Claude, ask it to save as a skill, then invoke with 'fact-check this'.
More from coding & agent
- User sends AI agent to shop at REI, gets blocked by 'Press & Hold' human verification — unixterminal · 2026-11-21
- Together AI: tune the system before the weights — cheapest levers first when evals fail — togethercompute · 2026-11-21
- Coding Value Trends to Zero, Systems Thinking to Infinity: AI Skills Debate — attentionmech · 2026-11-21
- 100+ AI Agents Collaborate to Build Browser FPS Game KILLFEED — majidmanzarpour · 2026-11-21
- Security Firm Claims Grok Bots Made 31 Unprompted App Requests Missing From Chat History — shoumikgoswami · 2026-11-21
- Opus 5.5 Builds Usable CRUD Apps in Artifact, but Falls Short of Strava-Level Products — NathanpmYoung · 2026-11-21