Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add ryanelian/crucible-agent-skill --skill cruciblegit clone --depth 1 https://github.com/ryanelian/crucible-agent-skillWrote 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/ryanelian/crucible-agent-skill/crucible)<a href="https://agentmods.dev/skills/ryanelian/crucible-agent-skill/crucible"><img src="https://agentmods.dev/badge/skills/ryanelian/crucible-agent-skill/crucible/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/ryanelian/crucible-agent-skill/crucible"><img src="https://agentmods.dev/badge/skills/ryanelian/crucible-agent-skill/crucible.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 | $0.00086 | $0.03509 |
| Opus 5 | $0.00043 | $0.01754 |
| Sonnet 5 | $0.00017 | $0.00702 |
| Haiku 4.5 | $0.00009 | $0.00351 |
Grade A, and why
crucible 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 9d 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 — 352 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Crucible
Mantra: Less code is more. Code that does not run cannot slow you down — prefer deletion and simpler paths so the app does less work per request.
Identity: You refine the scoped edit set (feature/refactor diff, or an explicit path) — delete, simplify, flatten. You do not add features, widen product scope, or redesign architecture.
Batching & done when
ACTIVE until done. Prove clean with an empty pass when you can; do not stop after one tidy-up because it “looks fine.”
Path scope: /crucible path/or/area (see Scope — explicit path overrides git).
10 passes per batch. N resets to 1 each batch. One Pass N: line per pass
(see Output). Never skip a number. Never start pass 11 in a batch. Never
auto-continue past the cap.
| Stop | Next |
|---|---|
| Empty pass (any N) | Done. No ask. |
| Pass 10 non-empty | Ask for another 10. |
| User yes | New batch, N = 1, same scope + lists. |
| User no / "stop crucible" / "normal mode" | Done. |
Loop in one turn when tools allow. Passes back-to-back until empty or cap;
no asks between passes. Yield mid-batch only if tools are blocked — then, if the
/loop skill is available, arm a wake (purpose crucible, next N + scope);
otherwise ask the user to continue. Never re-arm after empty pass or declined
batch. Do not invent a loop mechanism if /loop is not installed.
Scope
Not the whole repo. Resolve the edit set as follows:
- Explicit path (wins): If the user names a file or folder
(
/crucible src/services/billing.tsx, “crucible on src/auth”, etc.), that path is the edit set — even when those files are git-clean / unmodified. For a folder, include source files under it (skipnode_modules/, build output,.env, etc.). Do not require a dirty diff. - Else dirty work:
git diff+git diff --staged+ untracked that belong to the change (same junk skips). - Else clean tree, feature committed:
git diff <default-branch>...HEAD(three dots). Resolve<default-branch>frommain,master,trunk,dev, ororigin/HEAD/ the remote default — whichever exists. - Else fallback: files touched this session that belong to the change.
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.
- 9d ago First seen · 352 lines · 86 tokens per session scan A bddb9d7bffae
crucible is a skill published in the GitHub repository ryanelian/crucible-agent-skill (4 stars, last pushed 1mo ago), licensed MIT. It adds 86 tokens to every session and 3,509 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 2026-08-31.
Other skills, from other repositories
sdd-tasks
Break an SDD change into implementation tasks. Trigger: orchestrator launches task planning for a change.
go-testing
Trigger: Go tests, go test coverage, Bubbletea teatest, golden files. Apply focused Go testing patterns.
pr-blocker-summarizer
Summarizes open pull requests into a blockers-first standup digest. Activates when the user asks to summarize open PRs, find blocked pull requests, generate a PR standup, or triage review backlog from a PR export.
review-work
Post-implementation gate review: run manual QA on the real surface yourself, then launch ONE gate reviewer (never a panel) to audit goal, constraints, code quality, security, missed context, and QA evidence. Use before a PR handoff or when the user explicitly asks to review completed work.
pr-reviewer
Reviews a Git branch as a pull request against the base branch (auto-resolved from origin/HEAD — usually main or master) using this repo's own rules — the CLAUDE.md constitution and the .claude/skills/ that apply to the changed files. Fetches the branch, switches to it, diffs it against the merge-base, routes the…
ijfw-cross-audit
Generate a cross-platform multi-model audit (Trident) on a diff, brief, or artifact. Trigger: 'cross audit', 'Trident', 'second opinion', 'check with other models', 'check with other AIs', 'cross-check this', 'get another perspective', /cross-audit.