本文へ移動
cccskills
無料GitHub で公開

version-release

Choose and apply the correct semantic version bump for this repository. Use for every user-visible release, before merge when a change set should ship as patch, minor, or major, and whenever package/plugin/desktop version metadata must stay synchronized.

インストール方法を見る

含まれるファイル(2)

  • SKILL.md8.5 KB
  • references/version-checklist.md1.6 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Version Release

Use Semantic Versioning as the repository-wide release rule:

  • Patch (X.Y.Z+1) for backward-compatible bug fixes, documentation-only changes, dependency/security maintenance, internal refactors, and small user-visible improvements that do not create a substantial new capability.
  • Minor (X.Y+1.0) for backward-compatible feature additions or meaningfully larger product capabilities, including new workflows, pages, integrations, API fields/routes, or major UX surfaces.
  • Major (X+1.0.0) for backward-incompatible behavior, removed/renamed public contracts, required migrations, or fundamental product/architecture changes.

When a change fits more than one category, use the highest applicable bump. When uncertain between adjacent categories, prefer the higher bump or ask the user before releasing. Never infer the bump from commit count, diff size, or elapsed time alone. When the user explicitly requests a concrete version, that instruction overrides automatic classification. State the override and synchronize the requested version exactly.

CCAM release workflow

  1. Read the current root package.json version and summarize why the change is patch, minor, or major.
  2. Update the root package.json and root lockfile.
  3. Mirror the shipping version in desktop/package.json and desktop/package-lock.json.
  4. Update the OpenAPI version example in server/openapi.js, then run npm run openapi:yaml.
  5. Update the deployment surface, which carries the release in many places: docker-compose.yml (ccam-dashboard and ccam-mcp image tags), deployments/helm/agent-monitor/Chart.yaml (version and appVersion), every deployments/kubernetes/** manifest (app.kubernetes.io/version labels, image: tags, kustomize newTag), and deployments/scripts/deploy.sh (the --tag example and the sed image substitution).
  6. Update the version shown in DEPLOYMENT.md and docs/DEPLOYMENT.md (the ccam-dashboard:<version> substitution example) and in CITATION.cff (version:).
  7. Update version-sensitive UI snapshots when the rendered release string changes (the dashboard renders UI build v<version>; regenerate with cd client && npx vitest run -u and confirm the diff is only the version line).
  8. Run npm run extensions:sync so every Claude/Codex plugin manifest and marketplace stays on the root release.
  9. Update release/version documentation only where the concrete version is intentionally shown.
  10. Create or reuse the open GitHub milestone named exactly v<version> for the new root version. Query all milestones before creating one. If an exact-title milestone already exists and is closed, stop and verify whether that version has already shipped instead of creating a duplicate.
  11. Identify the current open pull request containing the bump and assign it to that milestone. Read closingIssuesReferences from the PR and assign every linked closing issue to the same milestone. If the branch has no PR yet, create/reuse the milestone now and treat PR/issue assignment as an incomplete release step until the PR exists.
  12. Verify the milestone on the PR and every linked issue with fresh GitHub reads. Do not infer completion from a successful edit command alone.
  13. Run npm run extensions:validate, relevant tests/builds, and ccam version or node bin/ccam.js version.
  14. Confirm only independently shipped packages remain on their own versions; do not bump client, mcp, monitoring, or VS Code extension packages unless those products are also being released.

The authoritative surface list

server/__tests__/release-version-consistency.test.js is the machine-checkable contract for a release bump. Read it first and run it last — it is more reliable than this document, because it fails when a surface drifts:

node --test server/__tests__/release-version-consistency.test.js

It asserts root/desktop packages and lockfiles (including packages[""].version), the live and generated OpenAPI versions, Compose and Helm metadata, Kubernetes labels/images/kustomize tags, every generated plugin manifest, both marketplace catalogs, the deployment guides and deploy.sh, and that the independently versioned subprojects have not been dragged along.

When you add a new file that carries the release version, add an assertion there in the same change. A surface with no assertion is a surface that will silently drift — CITATION.cff sat at 1.1.0 for many releases for exactly this reason.

Cross-check with a repo-wide sweep for the previous version before finishing:

previous_version="$(git show HEAD:package.json | node -p 'JSON.parse(require("fs").readFileSync(0,"utf8")).version')"
grep -rFn "$previous_version" --exclude-dir=node_modules --exclude-dir=.git \
  --exclude-dir=dist --exclude-dir=.worktrees .

grep -F matters: the dots in a version are regex wildcards otherwise, so a plain grep -r also matches unrelated strings like 2a0b11. Reading the previous value from HEAD:package.json keeps the command runnable as-is mid-bump, before the change is committed.

Expect zero hits outside package-lock.json history and deliberate historical references (for example "pre-v2.0.9 inflation" in the token-repair docs), which must stay pointing at the release they describe.

Not release-tied — do not bump

  • sw.js and client/public/sw.js cache names (landing-v2, dashboard-v2) are cache generations, not release versions. wiki/sw.js is bumped when wiki content changes, which is the update-project-docs skill's job, not this one.
  • deployments/helm/agent-monitor/values.yaml uses tag: "" and falls back to the chart's appVersion. Leave it empty.
  • client, mcp, monitoring, and vscode-extension ship on their own versions and are asserted not to equal the root release.

GitHub milestone workflow

Use the repository resolved from the current checkout and the intended authenticated GitHub identity. Run gh auth status before any mutation and stop if the active account is not the account intended for the repository:

repo="$(gh repo view --json nameWithOwner --jq .nameWithOwner)"
version="$(node -p "require('./package.json').version")"
milestone="v${version}"
pr="$(gh pr view --json number --jq .number)"

existing="$(
  gh api "repos/${repo}/milestones?state=all&per_page=100" --paginate \
    --jq ".[] | select(.title == \"${milestone}\") | [.number, .state] | @tsv"
)"
  • If existing is empty, create the milestone with gh api --method POST "repos/${repo}/milestones" -f title="${milestone}".
  • If it reports one open milestone, reuse it.
  • If it reports more than one match or a closed match, stop and resolve the repository state. Do not create another milestone with the same release title.
  • Assign the PR with gh pr edit "$pr" --milestone "$milestone".
  • Read linked issues with gh pr view "$pr" --json closingIssuesReferences. Assign each same-repository issue with gh issue edit <number> --milestone "$milestone".
  • If a closing issue belongs to another repository, stop and report it. Milestones are repository-scoped, so do not silently create or reuse a similarly named milestone in another repository.
  • Verify with gh pr view "$pr" --json milestone,closingIssuesReferences, then query each linked issue's milestone.

Release guardrails

  • Do not hand-edit generated Codex metadata or marketplace files after extensions:sync.
  • Do not create or move a Git tag unless the user explicitly requested a release/tag operation.
  • Creating the matching GitHub milestone and assigning the current PR plus linked closing issues is part of a version bump, not a release/tag operation.
  • Do not create duplicate milestones, guess issue relationships from prose when closingIssuesReferences is available, or leave linked release work assigned to a different version.
  • Do not call a breaking change “minor” merely because compatibility can be restored later.
  • Do not leave root, desktop, OpenAPI, snapshots, deployment manifests, or generated plugin versions out of sync.
  • Do not blanket find-and-replace the version across the repo: it would pull the independently versioned subprojects along and rewrite deliberate historical references.

References

  • Repository release checklist: references/version-checklist.md

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Inspect fired CCAM alerts and manage alert rules for token thresholds, event patterns, inactivity, and status duration. Use when acknowledging alerts, creating or editing a rule, checking cooldowns, or connecting alert rules to webhook targets.

日本語の概要は準備中です。原文の説明を表示しています。

hoangsonww/Claude-Code-Agent-Monitor1,0692026年10月12日 更新

Identify anomalous sessions using Agent Monitor data — cost outliers from the pricing engine, token anomalies (cache miss spikes, compaction baseline surges), unusual event type ratios (PreToolUse/PostToolUse gaps, APIError clusters), behavioral deviations from workflow intelligence (complexity score outliers, error propagation anomalies), and sessions with abnormal metadata (extreme turn_count, high thinking_blocks, zero turn_duration).

日本語の概要は準備中です。原文の説明を表示しています。

hoangsonww/Claude-Code-Agent-Monitor1,0692026年10月12日 更新

Produce a detailed report on APIError events from Agent Monitor data — counts over time, which sessions and models are affected, and the likely root cause (rate limits, overload/529, or context-window pressure) inferred from each event's summary and data payload. Use when API errors spike or when you need to explain why requests are failing.

日本語の概要は準備中です。原文の説明を表示しています。

hoangsonww/Claude-Code-Agent-Monitor1,0692026年10月12日 更新

benchmark

無料

Benchmark one session (or a small recent set) against the rolling average using Agent Monitor data — cost, total tokens, tool count, and workflow complexity score — and report where each metric lands as a percentile of the population. Tells you whether a session was normal, cheap, or an outlier. Use when judging whether a session was typical or out of band.

日本語の概要は準備中です。原文の説明を表示しています。

hoangsonww/Claude-Code-Agent-Monitor1,0692026年10月12日 更新

Define a spend budget for Claude Code and, optionally, create a cost alert rule that fires when usage crosses the limit, via POST /api/alerts/rules on the Agent Monitor dashboard. Reads current spend from /api/pricing/cost to size the budget sensibly and explains every rule field before writing. Use when setting a spend limit or wiring up a budget guardrail.

日本語の概要は準備中です。原文の説明を表示しています。

hoangsonww/Claude-Code-Agent-Monitor1,0692026年10月12日 更新

Analyze prompt-cache effectiveness for Claude Code usage from the Agent Monitor dashboard — cache hit rate (total_cache_read / (total_cache_read + total_input)), cache_write vs cache_read reuse, cache-read vs cache-write spend, and the sessions with the poorest reuse. Pulls token totals from /api/analytics, per-session detail from /api/sessions, and dollar splits from /api/pricing/cost. Use when diagnosing cache spend or deciding whether prompt caching is paying off.

日本語の概要は準備中です。原文の説明を表示しています。

hoangsonww/Claude-Code-Agent-Monitor1,0692026年10月12日 更新

hoangsonww のスキルをすべて見る

このスキルの問題を報告する