bayesian-reasoning-calibration — Claude Code skill
The bayesian-reasoning-calibration skill applies Bayes' theorem to systematically revise probability estimates as new evidence arrives. It helps you forecast more accurately, avoid overconfidence, and justify why your beliefs change.
no license 162 installs ★ 203 updated 2026-09-01
npx skills add lyndonkl/claude --skill bayesian-reasoning-calibration -a claude-codeWhat bayesian-reasoning-calibration does
The skill works in five stages: define the question, set a starting probability based on base rates and similar cases, assess how likely your data would be under each hypothesis, recalculate your belief, and check whether your reasoning holds up. It includes a calculator for straightforward cases and a framework for comparing multiple hypotheses at once. The guide provides worked examples (product launch, medical diagnosis) and a checklist to keep you on track.
The key constraint is honest initial probability estimates and realistic assessments of how likely your observations would be if each hypothesis were true. If you start biased or cherry-pick convenient facts, the output is unreliable. The skill works best for forecasting, risk assessment under uncertainty, and hypothesis testing.
You get a structured output document that records each step of your reasoning. This makes it easy to spot where assumptions might be wrong and to calibrate by comparing past predictions to actual outcomes later.
What it helps with
- Revise probability estimate given new evidence or data
- Calibrate confidence in a forecast and catch overconfidence
- Quantify how strongly new information should shift a decision
- Structure reasoning under uncertainty for team discussion
- Compare competing hypotheses and rank them by likelihood
Example prompts for the agent
- Should we ship this feature—what's the probability of success given beta test results?
- Assess server failure risk based on new error logs and metrics
- Update revenue forecast after a competitor's earnings report
How to install bayesian-reasoning-calibration
1Skills CLI (Vercel)
npx skills add lyndonkl/claude --skill bayesian-reasoning-calibration -a claude-code2Manually: put the files in place
~/.claude/skills/bayesian-reasoning-calibration/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/bayesian-reasoning-calibration/
1Skills CLI (Vercel)
npx skills add lyndonkl/claude --skill bayesian-reasoning-calibration -a codex2Manually: put the files in place
.agents/skills/bayesian-reasoning-calibration/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add lyndonkl/claude --skill bayesian-reasoning-calibration -a cursor2Manually: put the files in place
.cursor/skills/bayesian-reasoning-calibration/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install lyndonkl/claude skills/bayesian-reasoning-calibration2Manually: put the files in place
.github/skills/bayesian-reasoning-calibration/SKILL.md1Skills CLI (Vercel)
npx skills add lyndonkl/claude --skill bayesian-reasoning-calibration -a gemini-cli2Platform CLI
gemini skills install https://github.com/lyndonkl/claude --path skills/bayesian-reasoning-calibrationManually: put the files in place
.devin/skills/bayesian-reasoning-calibration/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add lyndonkl/claude --skill bayesian-reasoning-calibration -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about bayesian-reasoning-calibration
How do I install bayesian-reasoning-calibration in Claude Code?
Run npx skills add lyndonkl/claude --skill bayesian-reasoning-calibration -a claude-code, then open ~/.claude/skills/bayesian-reasoning-calibration/SKILL.md.
What does bayesian-reasoning-calibration do?
The skill helps you systematically revise your belief in a hypothesis as new evidence arrives. You set a starting probability, estimate how likely your data would be under each hypothesis, and the skill applies Bayes' theorem to compute the updated probability.
Why does a starting probability matter if new data will arrive anyway?
A starting probability (prior) anchors your reasoning and prevents you from overweighting recent data. It's grounded in base rates and reference cases, so it protects against the bias of ignoring background information.