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 igapyon/igapyon-agent-skills --skill igapyon-miku-scmgit clone --depth 1 https://github.com/igapyon/igapyon-agent-skillsWrote 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/igapyon/igapyon-agent-skills/igapyon-miku-scm)<a href="https://agentmods.dev/skills/igapyon/igapyon-agent-skills/igapyon-miku-scm"><img src="https://agentmods.dev/badge/skills/igapyon/igapyon-agent-skills/igapyon-miku-scm/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/igapyon/igapyon-agent-skills/igapyon-miku-scm"><img src="https://agentmods.dev/badge/skills/igapyon/igapyon-agent-skills/igapyon-miku-scm.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 | $0.00203 | $0.02752 |
| Opus 5 | $0.00102 | $0.01376 |
| Sonnet 5 | $0.00041 | $0.00550 |
| Haiku 4.5 | $0.00020 | $0.00275 |
Grade A, and why
igapyon-miku-scm 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 11d 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 — 191 lines — stays where its author put it; the contents beside it link to each section on GitHub.
igapyon-miku-scm
Run documented miku-soft source-control workflows with a small runtime context.
Detailed policy remains under references/ for design, maintenance, legacy
operations, and exceptions; migrated workflows execute through the fixed runner.
Current Scope
Support repository status and maintenance; PR, Release, About, and Issue work; recommit, publication, backup, and post-merge work; version and release audits; and deterministic READONLY, preflight, and apply workflows.
Activation-Only Fast Path
If no concrete SCM task is given, acknowledge activation and ask what is wanted. Do not inspect the repository or load detailed references yet.
Runtime Kernel
- Identify the exact repository and classify the request with scripts/miku-scm-workflow-manifest.mjs.
- For an ID in that manifest, invoke
scripts/miku-scm-run.mjs directly with that ID
and fixed options. Resolve that runner path relative to the directory
containing this
SKILL.md; never derive it from the target repository name or--repo, which identifies only the workflow target. Do not read its detailed Markdown during normal execution. Prefer human output for a mechanical workflow when no structured field is needed for a subsequent step. Because--formatis a runner-global option, always place it before the workflow ID:node <skill-root>/scripts/miku-scm-run.mjs --format human <workflow-id> [workflow options]. Never place--formatafter the workflow ID. Return the runner'shuman_outputunchanged instead of paraphrasing it. - Treat a bare
recommitrequest as the PR Soft Reset Recommit workflow with an implicit PR-writing request. The user does not need to sayPRseparately. Resolve the base first and draft from exactly<base>..HEAD. If no matching saved PR draft exists, collectwriting.pr.prepareevidence, draft and save the PR text, then continue topr.recommit.preflight. For an ordinary PR request without an explicit target, prefer the complete branch range when the branch is two or more commits ahead of its resolved base; keep a one-commit branch as a single-commit PR. Explicit commits and ranges always win. This routing authorizes preparation only; preserve the explicit approval gate before the local history rewrite. Treat exactpr recommit pushas the separate remote workflow. First run one READONLYpr.recommit.preflight. If it has any blocker other thanPR draft is unresolved or missing, return that fixed stop without drafting or mutation. If it has a matching draft and no blocker, invokepr.recommit.push --base <resolved-base> --pr-draft <resolved-draft> --applyonce. If the missing draft is its only blocker, runwriting.pr.prepare --target <resolved-base>..HEADonce, draft once from that evidence, save it at the returnedsuggested_draft_path, then invoke the same fixed push workflow with that exact base and saved draft in the same user turn. The runner never invents PR prose; this Skill-only writing fallback supplies the reviewed artifact it requires. Its exactpushtoken authorizes this bounded draft preparation and the fixed recommit-and-publication transition. Do not route barerecommitorpr recommitto it. Route exactmiku-scm git add committowork.commit --apply. When the current user request or current TODO heading identifies the work, pass that concise one-line title with--message; do not use the generic fallback. Omit--messageonly when no concrete title is available, or for a version-only change so the runner selects its deterministic version title. Then let the fixed runner stage all ordinary non-ignored changes, check, and commit without returning between successful steps. Do not reconstruct that Git command sequence. - For a current Issue status, backlog, or progress request, invoke
github.issue.read --listonce with its default Open state. Do not add a second--state allread merely to calculate an Open/Closed breakdown. Use--state closedor--state allonly when the user explicitly needs closed or historical Issues, completion metrics, or comparison with a closed Issue. - Never issue
ghdirectly. GitHub CLI access is allowed only inside the fixed bundled helpers. If no fixed helper or anonymous REST route exists, stop instead of inventing a command. - A READONLY result or preflight does not authorize mutation. Local rewrite, tracked-content change, remote mutation, deletion, tag change, Release publication, and version change each require the matching explicit request and workflow gate. Approval never transfers between workflows.
- Before tracked-content mutation or an ordinary commit, inspect the current
branch and worktree. Do not mutate a
-donebranch. Preserve unrelated changes. Operational drafts and workflows with their own branch checks keep those documented rules. - A standalone apply must consume the unchanged reviewed artifact and expected digest when its workflow defines one. In an exact ordered batch, a later mutation for the same Issue may use the fixed preflight helper to refresh only the snapshot expectations allowlisted by the batch contract after an earlier approved step succeeds. Its repository, Issue, operation, draft and digest, requested labels, close reason, duplicate target, and workflow contract must remain byte-for-byte equivalent. Never retry an unresolved mutation automatically.
- Report inspected, changed, and pending work. Keep tag recommendation separate from tag mutation; normal tag handoff is GitHub's Release UI.
- Route an exact
miku-scm pendingrequest togithub.issue.handoff.list. After a reviewed Issue preflight, an exactmiku-scm approverequest routes togithub.issue.handoff.apply --applyand still requires exactly one pending handoff. The legacymiku-scm 承認input remains accepted. - When the human supplies either a full handoff ID or the exact 12-character
approval suffix shown by the preflight or pending list, route
miku-scm approve <selector>togithub.issue.handoff.apply --handoff <selector> --apply, or routemiku-scm dismiss <selector>togithub.issue.handoff.dismiss --handoff <selector> --apply. The fixed helper resolves a suffix only when exactly one pending handoff in the target repository matches it; otherwise it stops. Never choose, invent, or reconstruct a selector or reviewed apply arguments. These exact commands are fast paths and do not require loading detailed references. - Route an exact
miku-scm approve batch <selector> <selector> [...]request togithub.issue.handoff.batch.apply, repeating--handoff <selector>in the exact human-supplied order and ending with--apply. Require two to twenty selectors that resolve to unique pending handoffs. Never infer membership or order fromall, recency, Issue numbers, or prior prose. The fixed workflow validates the complete batch before the first mutation. For later mutations to the same Issue, the fixed workflow runs a handoff-free dependency preflight after each preceding success and accepts only allowlisted snapshot-expectation changes caused by the ordered sequence. Any semantic change, failed refresh,not-applied,conflict, orunresolvedresult stops every later handoff. Preserve a delegateconflictas a distinct terminal state.
What ships with it
60 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- agents/openai.yaml 235 B
- docs/prompt-runner-ai-involvement-reduction-pattern.md 9.6 KB
- docs/work-cycle-lifecycle-redesign.md 47 KB
- index.json 7.3 KB
- references/approval-handoff.md 5.9 KB
- references/deterministic-workflow-runner.md 15 KB
- references/github-about-writing.md 1.1 KB
- references/github-anonymous-readonly.md 6.2 KB
- references/github-backup-branch.md 2.4 KB
- references/github-branch-status.md 3.6 KB
- references/github-cli-static-helper-policy.md 4.7 KB
- references/github-issue-close.md 3.0 KB
- references/github-issue-comment.md 3.1 KB
- references/github-issue-create.md 8.6 KB
- references/github-issue-label-update.md 2.9 KB
- references/github-issue-rewrite-handoff.md 7.4 KB
- references/github-issue-update.md 7.7 KB
- references/github-post-push-pr-url.md 3.0 KB
- references/github-post-recommit-publish.md 7.9 KB
- references/github-pr-recommit-push.md 4.0 KB
- references/github-pr-soft-reset-recommit.md 10 KB
- references/github-pr-writing.md 5.1 KB
- references/github-readonly-cache.md 975 B
- references/github-release-tag-handoff.md 2.8 KB
- references/github-release-writing.md 3.5 KB
- references/github-repository-url.md 1.3 KB
- references/github-writing-rules.md 6.9 KB
- references/local-git-readonly.md 3.1 KB
- references/local-git-snapshot.md 1.1 KB
- references/performance-benchmark.md 2.7 KB
- references/repository-maintenance.md 7.1 KB
- references/repository-precommit-checks.md 3.1 KB
- references/runtime-and-test-suites.md 2.4 KB
- references/safety-cost-and-errors.md 2.5 KB
- references/scm-rules.md 32 KB
- references/version-increment-confirmation.md 6.0 KB
- references/version-increment.md 6.2 KB
- references/version-tag-release-audit.md 6.1 KB
- references/version-workflow-runner.md 3.3 KB
- references/work-commit.md 4.6 KB
- references/workflow-contracts.md 11 KB
- references/workflow-routing.md 1.6 KB
- references/writing-mode.md 2.9 KB
- scripts/github-issue-close.mjs 18 KB runs code
- scripts/github-issue-comment.mjs 18 KB runs code
- scripts/github-issue-create.mjs 25 KB runs code
- scripts/github-issue-label-update.mjs 18 KB runs code
- scripts/github-issue-read.mjs 4.8 KB runs code
- scripts/github-issue-update.mjs 29 KB runs code
- scripts/github-issues-cache.mjs 5.8 KB runs code
- scripts/miku-scm-benchmark.mjs 13 KB runs code
- scripts/miku-scm-cli-contracts.mjs 27 KB runs code
- scripts/miku-scm-error-report.mjs 2.3 KB runs code
- scripts/miku-scm-fixed-command-runner.mjs 1.7 KB runs code
- scripts/miku-scm-github-readonly.mjs 5.9 KB runs code
- scripts/miku-scm-handoff.mjs 28 KB runs code
- scripts/miku-scm-help.mjs 13 KB runs code
- scripts/miku-scm-human-output.mjs 23 KB runs code
- scripts/miku-scm-jst-time.mjs 1.5 KB runs code
- scripts/miku-scm-local-snapshot.mjs 5.9 KB runs code
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.
- 11d ago First seen · 191 lines · 203 tokens per session scan A 7490b8fc3cf6
igapyon-miku-scm is a skill published in the GitHub repository igapyon/igapyon-agent-skills (2 stars, last pushed 26d ago), licensed Apache-2.0. It adds 203 tokens to every session and 2,752 once invoked, about $0.0010 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
release-version
Steps to cut a VNote version release (e.g. v4.3.0) — bump the version with scripts/updateversion.py, refresh and fill zhCN/ja translations via lupdate, write the changes.md changelog, and trigger the CI [Release] build. Use when asked to "release", "cut a release", "bump the version", or "prepare vX.Y.Z".
agent-release-swarm
Agent skill for release-swarm - invoke with $agent-release-swarm.
release-announcement
Write a release announcement — changelog, blog post, in-app note, or social post — that leads with user impact, names the audience, and includes upgrade/migration steps without filler.
multi-agent-release-manager
Cleans up the workspace, formats code, runs presubmit checks, and uploads CLs to Gerrit.
release-notes
Generate user-facing release notes from tickets, PRDs, or changelogs. Creates clear, engaging summaries organized by category (new features, improvements, fixes). Use when writing release notes, creating changelogs, announcing product updates, or summarizing what shipped.
pack-submit
Package one of this agent's own skills as a standalone community pack and submit it to the aeon registry as a PR.