verify
Verify any change, fix, behavior, deploy, or external state with current evidence from its real surface. Use when the user asks to verify, prove, check, test, or demonstrate that something works.
Install
npx skills add https://github.com/Melvynx/ai-maxxing-skills/tree/main/skills/verify
claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install melvynx-ai-maxxing-skills@llmmart
git clone https://github.com/Melvynx/ai-maxxing-skills.git
The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole melvynx/ai-maxxing-skills collection as a plugin from our marketplace. Git is the plain clone.
Skill manifest
Prove the claim (or, when none is given, the latest user request) from current evidence on its real surface.
- Define observable pass criteria.
- Exercise the real user-facing or authoritative surface: open the page in a browser, call the endpoint, run the command, read back the provider (Stripe, Convex, Cloudflare, Lumail). Static checks (typecheck, lint, build, tests) support runtime proof; they do not replace it.
- Capture current evidence:
- Visual: a screenshot of the page or state, shown in the chat.
- Non-visual: the command or request, its output, errors, status, and the authoritative read-back.
- Mark each criterion
PASSorNOT PROVEN.
Keep these proof layers distinct: local or static checks, provider read-back, deployed or public artifact, authenticated live behavior.
After any later change, rerun the affected criteria and replace the evidence.
Finish with PASS only when every criterion has current evidence. Otherwise BLOCKED: NOT PROVEN with the exact blocker.
Files (ai-maxxing-skills)
-
SKILL.md 1.3 KB
--- name: verify description: Verify any change, fix, behavior, deploy, or external state with current evidence from its real surface. Use when the user asks to verify, prove, check, test, or demonstrate that something works. argument-hint: "<thing to verify or expected outcome>" --- Prove the claim (or, when none is given, the latest user request) from current evidence on its real surface. 1. Define observable pass criteria. 2. Exercise the real user-facing or authoritative surface: open the page in a browser, call the endpoint, run the command, read back the provider (Stripe, Convex, Cloudflare, Lumail). Static checks (typecheck, lint, build, tests) support runtime proof; they do not replace it. 3. Capture current evidence: - Visual: a screenshot of the page or state, shown in the chat. - Non-visual: the command or request, its output, errors, status, and the authoritative read-back. 4. Mark each criterion `PASS` or `NOT PROVEN`. Keep these proof layers distinct: local or static checks, provider read-back, deployed or public artifact, authenticated live behavior. After any later change, rerun the affected criteria and replace the evidence. Finish with `PASS` only when every criterion has current evidence. Otherwise `BLOCKED: NOT PROVEN` with the exact blocker.
Comments (0)
Sign in to join the conversation.
Reviews (0)
No reviews yet.
No comments yet.