Borrowing it
Nothing to install: this file belongs to radekamirko/C.R.I.S.P. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/radekamirko/C.R.I.S.P/main/.claude/skills/phase5-prove/SKILL.mdgit clone --depth 1 https://github.com/radekamirko/C.R.I.S.PWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/radekamirko/c.r.i.s.p/phase5-prove)<a href="https://agentmods.dev/skills/radekamirko/c.r.i.s.p/phase5-prove"><img src="https://agentmods.dev/badge/skills/radekamirko/c.r.i.s.p/phase5-prove/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/radekamirko/c.r.i.s.p/phase5-prove"><img src="https://agentmods.dev/badge/skills/radekamirko/c.r.i.s.p/phase5-prove.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00074 | $0.01045 |
| Opus 5 | $0.00037 | $0.00522 |
| Sonnet 5 | $0.00015 | $0.00209 |
| Haiku 4.5 | $0.00007 | $0.00104 |
Grade A, and why
phase5-prove scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 93 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Pacing Rule — One Section at a Time
Present one section, deliverable, or elicitation move at a time. After presenting — stop. Wait for the client to respond. Do not pre-fill and present multiple sections in one message. Do not move to the next step until the client confirms, corrects, or gives a clear go-ahead.
The goal is a conversation, not a document dump. If you've written more than one section without a client response in between — you've gone too far.
P — Prove: Success Validation
Did the needle move?
Project State
At the start of Phase P: read
docs/crisp-state.json. Pullphases.R.baselineandphases.R.successTargetas the measurement baseline. Checkphases.S.sprintsto know what was built.At the end of Phase P: update
docs/crisp-state.json:
- Add
"P"tophases.complete- Set
phases.P.completetotrue- Set
phases.P.outcometo"success","partial", or"fail"- Add summary notes to
phases.P.notes
This phase closes the loop back to Phase R. Three questions drive everything.
1. Did the needle move?
Pull the baseline and targets from docs/success-metrics.md and measure against them:
- Quantitative: Did each metric hit its target? (e.g. "quotes sent in 4min vs 4h before")
- Qualitative: Re-run the survey or interview from the Phase R baseline frustration score — did it improve?
- Second-order effects: Did the downstream impacts materialise? (e.g. cart abandonment dropped after checkout speed improved)
2. Was everything actually built and verified?
Read docs/test-log.md. For every sprint that shipped:
- Is there a test log entry?
- Did every specced behaviour (from the sprint's AI Spec test requirements) get a test written and run?
- Are there any ❌ failures that were never resolved?
A system where the metrics look good but tests weren't written is a system running on luck. The test log is how you know the difference between "it works" and "it worked that one time we checked."
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 12d ago First seen · 93 lines · 0 tokens per session scan A e28a2a21b52f
phase5-prove is a skill published in the GitHub repository radekamirko/C.R.I.S.P (52 stars, last pushed 3mo ago), licensed MIT. It adds 74 tokens to every session and 1,045 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
janitor-discover
Find new skills on GitHub or check a specific skill before installing. Use when the user wants to search for skills, evaluate a skill URL, check overlap and security risk before installing, or compare a local skill against alternatives. Trigger with '/janitor-discover'.
alibaba-maestro
Alibaba Cloud Maestro routing skill. Classify the user's Alibaba Cloud task, select the narrowest specialist agent or the right team of specialists from the catalog, and dispatch them — single specialist for focused tasks, parallel team (max 4) for multi-domain tasks. Never auto-dispatch live-guard agents.…
azure-maestro
Use this skill to classify a user task, select the right Azure specialist agent or team of specialists from the catalog, and dispatch them. Single specialist for focused single-domain tasks; parallel team (max 4) for tasks that span multiple domains. Never auto-dispatches live-guard agents — those always pause for…
accounting-maestro
Route accounting questions to the narrowest specialist in the catalog. Use when you do not already know the specialist needed. Not for direct accounting answers; Maestro classifies, dispatches, and synthesizes only. Dispatches single agent for focused tasks, parallel team (max 3) for multi-domain tasks. Never…
model-registry-refresh
Re-verify and extend catalog/model-registry.json — the fail-closed model-name and reasoning-effort matrix scripts/model-policy.mjs validates against — via delegated Context7-backed research, orchestrator-owned registry edits, and the full validation chain; use when a policy check fails on an unregistered model, a…
add-educational-comments
Add educational comments to code files to transform them into effective learning resources. Explains the "why" behind syntax, idioms, and design choices, aligned with the learner's knowledge level and educational goals.