BUZZARDGTA/Session-Sniffer / .agents/skills/release
release
VerifiedPrepare, commit, and push a Session Sniffer release by updating the version in pyproject.toml with the current UTC build timestamp, incrementing the RC or final version as requested, synchronizing uv.lock via uv lock, validating the change, creating the release commit, pushing the branch to GitHub, and generating a synthesized user-facing changelog since the previous release.
Session Sniffer Release
Use this skill when the user wants to prepare and commit a new Session Sniffer release or release candidate.
Release Procedure
- Inspect the Git working tree before making changes.
- Run the release helper script in check mode to inspect
pyproject.toml, published GitHub tags, and the current UTC timestamp:python .agents/skills/release/scripts/prepare_release.py --check(See prepare_release.py.) - Check whether the current version has already been published/tagged on GitHub (
is_published_on_githubfromgit ls-remote --tags originor local tags matching the version prefix):- If NOT tagged/published on GitHub: do NOT increment the RC or version number; only update the build timestamp.
- If ALREADY tagged/published on GitHub: ask the user whether this is a new release candidate (
RC) or a final release.
- Run prepare_release.py to generate the current UTC timestamp (
YYYYMMDD.HHMM/YYYY-MM-DDTHH:MM:SSZ), updateversioninpyproject.toml, runuv lock, enforce CRLF (\r\n) line endings on bothpyproject.tomlanduv.lock, and verify synchronization viauv lock --check:- If unpublished:
python .agents/skills/release/scripts/prepare_release.py --mode timestamp(preserves the existing version prefix, e.g.v1.5.0rc.69, and updates only+YYYYMMDD.HHMM). - For a new RC:
python .agents/skills/release/scripts/prepare_release.py --mode rc(increments the existingrc.Nnumber by one, preservingX.Y.Z, and appends+YYYYMMDD.HHMM). - For a final release:
python .agents/skills/release/scripts/prepare_release.py --mode final(removes therc.Nsuffix, preservingX.Y.Z, and appends+YYYYMMDD.HHMM).
- If unpublished:
- Review the resulting diff (both
pyproject.tomlanduv.lockshould reflect the new version). - Run the release-specific validation before committing (see Validation below).
- Create exactly one version-bump commit containing both
pyproject.tomlanduv.lock(see Commit below). - Push the current branch to GitHub (
git push) immediately after creating the version-bump commit sooriginalready contains the updatedpyproject.tomlanduv.lockbefore the user publishes the GitHub release/tag. The release tag MUST matchversioninpyproject.tomlexactly. - Show the user:
- previous version,
- new version,
- changed files (
pyproject.toml,uv.lock), - validation performed and confirmation that the branch was pushed to GitHub,
- a copy-ready GitHub Release Tag in its own fenced code block (the exact
versionfrompyproject.toml, e.g.v1.5.0rc.93+20261027.2106— never the PEP 440-normalizeduv.lockversion1.5.0rc93+...), - a copy-ready GitHub Release Title in its own fenced code block (
<version-without-build-metadata> - YYYY/MM/DD (HH:MM)), - a synthesized user-facing changelog from the previous release to
HEAD(following the User Changelog rules below).
User Changelog
Always include a concise, well-structured user-facing changelog covering all net changes between the previous published release tag (or previous build: bump version commit) and HEAD.
Synthesizing the Net Final State (Diff-First, Never Commit-Message-First)
Do not mechanically list, paraphrase, or trust commit subject lines (git log). Commit messages frequently describe intermediate designs, sub-features that were later removed before HEAD, or internal fixes to unpublished code from earlier in the same cycle.
Instead, follow a strict diff-first workflow:
- Inspect the net code diff (
git diff <prev-release>..HEAD -- src/ resources/) and the actual code atHEADas the sole source of truth for what changed compared to<prev-release>. - Use
git log <prev-release>..HEADonly to attribute commit SHAs to the net changes you have already verified ingit diff <prev-release>..HEAD.
Apply these mandatory rules to every entry:
- Verify every sub-item of a commit against
git diff <prev-release>..HEAD: When a commit message mentions multiple items (e.g."add visual indicators, ON/OFF toggle pill, and context menu toggle"), inspectgit diff <prev-release>..HEADto confirm each item still exists inHEAD. If one part (such as a context menu toggle) was later removed or reworked in a subsequent commit beforeHEAD, omit that removed part completely—never copy it from the earlier commit's subject line. - Consolidate multi-commit work without leaking intermediate fix details: If a feature or change was introduced after
<prev-release>and later refined, fixed, guarded, or tweaked in subsequent commits during the same release cycle, merge all contributing commit links into a single cohesive entry describing only the final user-facing behavior. Do not append internal stabilization clauses or intermediate commit-subject phrasing (e.g."with automatic operator flipping...","concurrent file-read retry handling...", or"guard unresolved...") to the changelog description. - Drop transient intermediate bugs (
### Fixedis strictly for bugs present in<prev-release>): Every entry under### Fixedmust fix a bug that existed in<prev-release>. If a commit (fix(...)) fixed a bug, regression, or edge case that was introduced after<prev-release>by an earlier commit in the same<prev-release>..HEADrange, never list it under### Fixedand never describe the transient bug—at most, include its commit link alongside the parent feature/change entry. - Drop reverted or removed experiments: If a feature, UI control, or refactor was added after
<prev-release>and later reverted or removed beforeHEAD, omit it completely when its net change vs.<prev-release>is zero. - Filter internal noise: Omit purely internal housekeeping commits (e.g.
.gitignoretweaks, agent/skill config edits, lockfile/dependency bumps, version bumps, type-annotation-only fixes, or pure code formatting) that have no user-visible behavior, stability, or performance impact. - Verify exact integrations, filenames, paths, and UI labels from
HEADandREADME.md: Never guess or rely on outdated/archived conventions (e.g. all settings and UserIP databases use.json, not.ini; cross-check active plugins/integrations againstREADME.mdandHEAD). Check the actual diff/source atHEADto cite exact file formats, log paths, and data sources. - Group by category: Organize entries under relevant headings such as
### Added,### Changed,### Fixed,### Performance, or### Removed(include only categories that have entries). - Clickable GitHub commit links: Format every referenced commit hash as a clickable Markdown link to the repository commit on GitHub:
[`<short-sha>`](https://github.com/BUZZARDGTA/Session-Sniffer/commit/<full-sha>). When a consolidated entry spans multiple commits, link each contributing commit (e.g.([`7c0bab5`](https://github.com/BUZZARDGTA/Session-Sniffer/commit/7c0bab5f80a2b7b824883b713801af53921d6303), [`6841708`](https://github.com/BUZZARDGTA/Session-Sniffer/commit/68417080625ab90e9fb92ae4902736c05b318e8d))). - Copy-ready raw Markdown block: Provide the changelog in a fenced ````markdown
code block so it can be copied in one click into the GitHub Release notes, where.github/workflows/build.yml(notify-discord`) automatically forwards it to the Discord release webhook once the Windows, Linux, and macOS binaries finish compiling.
Pre-Output Changelog Audit Checklist
Before presenting the changelog to the user, audit every bullet and sub-clause against these three checks:
- Net-Diff Check: Does every feature, UI element, or behavior mentioned in this sentence actually exist in
git diff <prev-release>..HEADatHEAD, or did it come from an earlier commit subject line that was later modified/removed? - Baseline Bug Check: For every item in
### Fixed, did that bug exist in<prev-release>? (If it was fixing code introduced after<prev-release>, remove it from### Fixed.) - No Intermediate Mechanics Check: Does any feature bullet include internal implementation/fix phrasing copied from follow-up
fix(...)commit subjects? (If so, strip the fix phrasing and keep only the clean user-facing description of the final feature.)
Version Format
The expected version format is:
vMAJOR.MINOR.PATCH+YYYYMMDD.HHMM
or for release candidates:
vMAJOR.MINOR.PATCHrc.N+YYYYMMDD.HHMM
Examples:
v1.5.0rc.45+20260909.2017
v1.5.0+20260909.2017
Unpublished / Untagged Releases
If the current version in pyproject.toml has not been published or tagged on GitHub (e.g. no remote tag exists matching the current version prefix, such as v1.5.0rc.69):
- Do NOT increment the version number or
rc.Nnumber. - Retain the exact existing version prefix.
- Only update the UTC build timestamp (
+YYYYMMDD.HHMM).
For example, if the current version is:
v1.5.0rc.69+20260929.0504
and rc.69 was never tagged or published on GitHub, the next build timestamp update becomes:
v1.5.0rc.69+<current UTC timestamp>
RC Releases
If the current version is:
v1.5.0rc.44+20260908.1641
the next RC must become:
v1.5.0rc.45+<current UTC timestamp>
Do not reuse the previous timestamp.
Final Releases
If the current version is:
v1.5.0rc.45+20260909.2017
a final release of version 1.5.0 becomes:
v1.5.0+<current UTC timestamp>
Do not retain the rc.N suffix.
If the user intends a different X.Y.Z final version, confirm that version before changing the file.
Commit
The commit subject must be:
build: bump version to <new-version>
For example:
build: bump version to v1.5.0rc.45+20260909.2017
The version-bump commit should contain only the intended release-version changes in pyproject.toml and uv.lock. Never commit a release version bump without its synchronized uv.lock.
Do not create a commit if validation fails.
Validation
For a version-only change, use the smallest relevant validation available:
- Run
uv lock --checkto ensureuv.lockis strictly up to date. - Verify CRLF (
\r\n) line endings on bothpyproject.tomlanduv.lock.
Do not run the full python code_quality_checks.py suite unless the release also includes significant code, dependency, resource, packaging, or other release-sensitive changes; in that case broaden validation appropriately.
Important
The version timestamp represents the current UTC time.
Do not use local time.
Do not manually invent or reuse a timestamp.
When /release is invoked, always push the branch to GitHub (git push) after creating the release commit so the remote branch is up to date before the GitHub release/tag is created.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-changelog-updatemorluto/reaPrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization. | Stars: 465k | Updated | Claude CodeCodex | Verified | Install |
| academy-guideanthropics/skillsStop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| claude-apianthropics/skillsReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file). | Stars: 196k | Updated | Claude CodeCodex | Unverified | Install |
| doc-coauthoringanthropics/skillsGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |