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 agentmods add skills/mickeyyaya/evolve-loop/commitnpx skills add mickeyyaya/evolve-loop --skill commitgit clone --depth 1 https://github.com/mickeyyaya/evolve-loopWrote 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/mickeyyaya/evolve-loop/commit)<a href="https://agentmods.dev/skills/mickeyyaya/evolve-loop/commit"><img src="https://agentmods.dev/badge/skills/mickeyyaya/evolve-loop/commit.svg" alt="Measured on agentmods" 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 | $0.00105 | $0.01656 |
| Opus 5 | $0.00053 | $0.00828 |
| Sonnet 5 | $0.00021 | $0.00331 |
| Haiku 4.5 | $0.00011 | $0.00166 |
Grade A, and why
commit 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 3d 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 — 62 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/evo:commit
Gated interactive commit. A commit only lands if simplify → review → lint → targeted-test all pass; the gate writes
.commit-gate/attestation.jsonbound tosha256(git diff HEAD). The commit goes throughevolve ship --class manual, which hard-verifies that attestation at this repo's real commit chokepoint — ship-gate forbids baregit commit, and ship commits internally, so this Go-side check (go/internal/phases/ship/commitgate.go) is the single enforcement point. This skill is the intelligent driver — the gate itself isevolve commit-gate run(Go;go/internal/commitgate/).
Procedure
Follow in order. Do not edit any file between step 4 and step 5 — that would make the attestation stale and ship will refuse.
-
Stage everything:
git add -A. The sanctioned commit path (evolve ship --class manual) commits the whole working tree, so this is an all-or-nothing commit (not a partial one). Staging now also lets new files be reviewed and bound into the attestation. -
Detect changed languages:
git diff --name-only HEAD, then map extensions (.go→go,.py→python,.ts/.tsx/.js/.jsx→ts,.rs→rust). -
Review chain — invoke via the Agent tool on the staged diff. Apply every CRITICAL/HIGH finding, re-stage (
git add -A), and re-run until clean. Record the names you actually ran. Two reviewers are required:# Agent (or skill) Covers a simplify: code-simplifier(orecc:code-simplifier)clarity, dead code, simplification b one review: the matching language reviewer ( go-reviewer/python-reviewer/typescript-reviewer/rust-reviewer) OR the generalcode-reviewer//code-review— pick onecorrectness, security, semantics Prefer the language reviewer when there's a clear primary language (richer); use the general
code-reviewerfor mixed/other languages. The combinedevo:code-review-simplifyskill satisfies both a and b in a single pass. -
Run the gate:
"$CLAUDE_PROJECT_DIR/go/bin/evolve" commit-gate run --reviewers "<comma-list of what you ran>".- This is the gate (Go;
go/internal/commitgate/): lang-detect, the--reviewersprecondition, the lint lanes, and the tree-SHA-bound attestation. (The original bash runner was deleted once a differential-parity test proved the two byte-identical; a Go-only golden test now pins the attestation byte layout.) - exit 0 → attestation written; proceed immediately to step 5 (no edits in between).
- exit 1 → reviewer-precondition or lint/test failure, or nothing staged (read stderr). Fix, re-stage, return to step 3 for the affected files.
- exit 2 → git/SHA fatal (not a repo,
git diff HEADfailed). Stop and report. - exit 3 → a required tool is missing/could not be installed; surface the printed manual install command to the user and stop.
- This is the gate (Go;
-
Commit + push via the sanctioned path (handles both, on the current branch):
"$CLAUDE_PROJECT_DIR/go/bin/evolve" ship --class manual "<message>"--class manualre-stages (git add -A, a no-op after step 1), verifies the commit-gate attestation matches the staged tree (the Go-side hard gate), commits to the current branch, and pushesorigin/<branch>.- Non-TTY agents (LLM invocations): the command will fail with a non-TTY error that includes the exact IPC auto-confirm env var to set. Set it before re-running. The signal name comes from
envShipAutoConfirmingo/internal/phases/ship/verify.go. - If it refuses with "stale" or "missing … attestation" → a file changed after step 4; return to step 3.
-
Watch CI (skip with a clear note if
ghis absent or the repo has no workflows):gh run watch "$(git rev-parse HEAD)"(orgh pr checks --watch).- green → done; report the run URL.
- red → perform one auto-fix pass:
gh run view --log-failed→ fix → repeat steps 1–5 as a new commit (never force-push). Re-watch once. If still red, stop and report the failing job + logs for the user to decide.
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.
- 3d ago First seen · 62 lines · 105 tokens per session scan A 58246f317e20
commit is a skill published in the GitHub repository mickeyyaya/evolve-loop (5 stars, last pushed 3d ago), licensed Apache-2.0. It adds 105 tokens to every session and 1,656 once invoked, about $0.0005 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 2026-09-03.
Other skills, from other repositories
review
Structured code review with parallel audit agents, confidence-scored triage, and optional auto-fix. Examines uncommitted changes, staged diffs, commit ranges, or specific paths. Produces a tiered report (MUST-FIX / RECOMMENDED / NIT) backed by evidence, then optionally applies fixes with verification.
pr-review
Address feedback left on a GitHub pull request: fetch unresolved review threads, make agreed Elixir/Phoenix code fixes, reply, and resolve. Use for a PR URL/number or reviewer comments. NOT for pre-PR review, findings triage, or CI monitoring.
phx-pr-review
Address feedback left on a GitHub pull request: fetch unresolved review threads, make agreed Elixir/Phoenix code fixes, reply, and resolve. Use for a PR URL/number or reviewer comments. NOT for pre-PR review, findings triage, or CI monitoring.
procoder
Work like a senior developer in a repository governed by procoder: run the commit gate before calling anything done, format and lint through the binary, and drive the spec, plan, todo, backlog, and sprint chain in .procoder/. Use this skill when the repository contains a .procoder/ directory or an AGENTS.md naming…
pr-feedback
Work a reviewer's comments on a pull request to the end - all three comment surfaces enumerated before any is triaged, every comment fixed, declined with a reason, or ticketed, push before you reply. Use when a PR comes back with review feedback or a red check.
pr
Pull request excellence and review optimization.