About bulletproof
bulletproof is a verification skill for assessing claims against evidence and preserving decisive implementation checks. It guides analysts through defining proof criteria, inspecting sources, and reporting claim-level findings.
Key features
- 01Define claims and proof criteria
- 02Inspect authoritative evidence
- 03Test absent or stale evidence
- 04Classify claims by evidence
- 05Record decisive implementation runs
Use cases
- 01Assess factual claims in a report
- 02Verify claims against local or primary sources
- 03Document command results after implementation
Install
Copy the command and run it in a terminal.
npx skills add https:/After installing, type/bulletproofin Claude Code to confirm it loaded.
Details
From the project READMEBulletproof
Trust nothing. Prove everything checkable.
Method
Run the matching branch. Run both only when claims and completed implementation both need proof.
Verification
- Freeze the claim set, target, and state.
- Expose every load-bearing assumption.
- Define proof and falsification before searching.
- Inspect authoritative local sources. Use primary sources for external facts.
- Run safe checks against absent, stale, or indirect evidence.
- Let evidence kill disagreement.
Implementation Proof
Preserve decisive command facts after code or validation work.
Output
For verification, return one rendered Markdown table:
| ID | Assessment | Evidence |
|---|
Give every identifiable claim one row:
Supported: decisive evidence establishes it.Unsupported: decisive evidence contradicts it.Unverified: available evidence cannot establish or contradict it.Blocked: required evidence exists but cannot be obtained.
Name the exact gap for Unverified and Blocked.
For implementation proof, return each decisive run once in chat: command, working directory, state identifier, numeric exit, and the smallest untouched decisive output excerpt.
Boundaries
Freeze the verification target. Acceptance and requirement conformance are outside this method.