SKILL.md
5,159 tokens · o200k_base · 23,666 bytes
OpenClaw Changelog Update
Use this for changelog rewrites and GitHub release-note source text. For regular
beta/stable, prepare complete notes before final-source qualification when
possible; Code SHA may then also be Release SHA. Editorial work may overlap
Code validation. If notes change afterward, a genuine CHANGELOG-only descendant
may use the existing product-evidence reuse policy. For
extended-stable, run it before final exact-head validation and tagging. Do not
rerun it for tooling retries, resumed publication, or promotion.
Use it with release-openclaw-maintainer; this skill owns changelog content,
ordering, grouping, and attribution discipline.
Goal
Rebuild the target CHANGELOG/YYYY.M.PATCH.md release section from a complete, generated
history manifest, not stale draft notes. Produce grouped user-facing release
notes sorted by user interest while preserving every relevant issue/PR ref and
every human Thanks @... attribution.
CHANGELOG.md is the generated release index. The shared owner
scripts/lib/release-changelog.mjs resolves release sections and contribution
records from a working tree or a pinned Git ref, including historical refs
that still contain a monolith. Use node scripts/release-changelog.mjs read --version <version> [--ref <sha-or-tag>] for a section, or add --record for
its contribution record. A missing split artifact must not fall back to the
root index. Historical releases without records gain no invented provenance.
Generated CHANGELOG/** files retain exact migrated or mirrored bytes and are
excluded from generic formatting, like the root changelog. Validate them through
their owner with pnpm changelog:check.
Current writes use the split layout. To save a complete initial section, use
node scripts/release-changelog.mjs write --version <version> --file <section.md>;
it updates the release entry, matching CHANGELOG/records/<version>.md when
present, and index together. Initial generation retains the section format
below. Published docs mirrors follow the separate post-release route at the end
of this skill; initial generation must never overwrite them.
Inputs
- Target base version:
YYYY.M.PATCH, without beta suffix. - Base tag: last reachable shipped release tag, usually the previous stable or the previous beta train requested by the operator. It must be an ancestor of the target; a newer but divergent tag is not a valid history boundary. Use an explicit shipped/main-closeout SHA only when it is also reachable from the target.
- Target ref: the exact product-complete history being documented. Its contribution-record target must be an ancestor of the final release target; it need not name a not-yet-created changelog commit. Include any later fixes before finalizing notes. Final notes may be committed before qualification, or afterward as a CHANGELOG-only descendant of a green Code SHA.
- Canonical main ref: current
origin/main, fetched before verification. Release notes cite the original merged main PR when the same work is carried by a backport. A release-branch PR is used only while no forward-port exists on current main.
Workflow
-
Confirm the release branch and exact history target:
git fetch --tags origin- confirm clean
git status -sb - record
git rev-parse HEADas the history target - record the Full Release Validation run id and attempt when qualification already exists
- finish pending product/version/backport changes before freezing final source; refresh the inventory for actual changes
-
Audit history, including direct commits:
git log --topo-order --date=iso-strict --pretty=format:'%h%x09%ad%x09%s' <base-tag>..<target-ref>git log --topo-order --grep='(#' --date=short --pretty=format:'%h%x09%ad%x09%s' <base-tag>..<target-ref>- Include every commit reachable from the target but not the base, including merged side branches. Resolve PR associations before deduplicating and subtracting shipped records; retain the existing revert exclusions.
- also inspect
--since='24 hours ago'when main moved during the release.
-
Generate the complete contribution record and editorial manifest before writing grouped prose:
node --import tsx .agents/skills/openclaw-changelog-update/scripts/verify-release-notes.mjs \ --base <base-tag> \ --target <target-ref> \ --main-ref origin/main \ --version <YYYY.M.PATCH> \ --manifest /tmp/openclaw-release-<YYYY.M.PATCH>.json \ --write-ledgerAdd repeatable
--release-provenance '<40sha> -> #PR[, #PR]'inputs when release commits cannot carry provenance metadata. These use the same exact marker grammar and current-main validation as commit-body markers.The verifier automatically reuses public GitHub GraphQL responses from an exact base/target SHA snapshot under the worktree's git metadata. Iterative rewrites at the same target avoid repeated network discovery. Use
--refresh-github-snapshotafter suspect API data,--github-snapshot <path>for an explicit artifact, or--no-github-snapshotfor a live-only audit. GitHub release bodies are always read live. ExplicitCI #,CI run #,Actions run #, andworkflow run #references in active source are classified separately only when issue/PR resolution fails and a live same-repository Actions lookup confirms the exact run ID. Any ordinary occurrence of that number in active source, notes, or the contribution record remains a strict issue/PR requirement. Confirmed runs appear asworkflowRunsin verification output and the manifest, never as PR associations or contributor credit.- the manifest is the required input to the rewrite, not an after-the-fact audit; it contains every referenced PR, eligible contributor credit, inline issue context, every direct commit, and an editorial-eligibility classification for PRs and direct commits
- schema version 3 is the required ephemeral manifest contract. Regenerate older manifests; version 2 is not a supported downstream-reader boundary
- for a historical backfill, add
--seed-ref <pre-backfill-ref>once so contribution records from the prior changelog are retained even when an older merged commit omitted its PR number; the verifier excludes records for work reverted after the base tag, including beta work reverted before the stable release - generated provenance reports in-range PRs separately from retained seed-only PRs, then states the unique row total. A PR present in both inventories counts as in-range; never describe the seed-inclusive total as work merged in the current release range
- add repeatable
--shipped-ref <prior-shipped-tag>when the reachable main closeout differs from the shipped tag or later forward-port commits re-associate PRs that were already released. Each tag is a cumulative shipped boundary: the verifier unions explicit PR rows from complete contribution records in numbered release sections, excludes only overlapping PRs, and ignoresUnreleased. Never infer this boundary from the base SHA, target prose, or target record. The manifest and generated provenance retain each tag plus the exact excluded PR inventory and count for deterministic candidate validation - source PR discovery bounds GitHub commit associations to the selected target history, or the frozen main history for canonical carriers. A contextual source reference becomes a contribution only when its merged commit is reachable in the target history and its merge time is within the target. Keep all references resolvable, but do not promote unrelated PRs merely because they merge while release preparation continues. Explicit seeds retain their historical membership and remain seed-only unless independently proven in-range. Resolve every association page; existing canonical, cherry-pick, and provenance contracts remain authoritative.
- explicit multi-commit reverts require a revert subject and one standalone
Reverts <full SHA> and <full SHA>.declaration (comma-separated lists with finalandalso work). The exact endingto restore the previous behavior.is accepted. Duplicate, abbreviated, embedded, or repeated declarations do not establish reversal. Each named commit must be a single-parent ancestor, and reverse-applying all named patches must reproduce the complete revert tree. Recognized declarations that fail this proof stop verification. Proof uses private Git index/object storage without hooks or external diffs; canonical single-revert and revert-of-revert accounting stays intact. - canonicalize backports to the original merged PR on
main: explicit cherry-pick origins win, then a unique normalized-subject match requires the same author and an overlapping changed path. Suppress release/backport PRs whenever the corresponding main PR exists on currentorigin/main. Keep a release-branch PR only when that change landed there first and has not yet been forward-ported tomain - read the manifest before editing
### Highlights,### Changes, or### Fixes; do not carry old grouped prose forward without re-auditing it - inspect linked PRs/issues or diffs for ambiguous commits. Direct commits are editorial input, not public ledger rows; infer material user outcomes from subject, body, touched files, tests, and nearby commits
-
Rewrite one stable-base section only:
- use
## YYYY.M.PATCH - do not create beta-specific headings
- do not leave a stale
## Unreleasedsection above the target release - if
Unreleasedcontains release-bound notes, fold them into the target section instead of deleting them
- use
-
Section shape:
### Highlights: 5-8 bullets, broad user wins first- include only a clear user-visible capability or workflow unlock, a material reliability/safety fix, a broad cross-surface improvement, or a release-defining integration/compatibility milestone
- every highlight must say what changed for a user in one sentence; use one user story per bullet and group its supporting PRs
- exclude tests, CI, refactors, docs, catalog churn, and implementation detail unless the outcome is a material install/update, data-safety, or widely visible user improvement
### Changes: new capabilities and behavior changes### Fixes: user-facing fixes first, grouped by impact and surface- group related changes/fixes by surface and user impact; avoid one bullet per tiny commit when several commits tell one user-facing story
### Complete contribution record: generated PR-first record after the grouped prose; it is the exhaustive accounting surface, not a second release summary
-
Preserve attribution:
- keep
#issue,(#PR),Fixes #..., andThanks @... - every human-authored merged PR represented by a user-facing entry needs
its PR ref and
Thanks @author, even when the PR had no linked issue - every human issue reporter for a
Fixes #...or referenced bug issue represented by a user-facing entry needsThanks @reporterunless the same handle is already thanked in that bullet - every human
Co-authored-bycontributor on represented user-facing work needsThanks @handlewhen a GitHub handle is known - when grouping multiple PRs/issues in one bullet, include every relevant PR/issue ref and every human contributor handle in that same bullet
- multiple
Thanks @...handles in one bullet are expected; do not drop or collapse contributor credit just because the note is grouped - if one grouped bullet covers both direct commits and PRs, keep all PR refs and thanks, plus any issue refs and human credit from the direct work
- issues remain normal inline
#NNNreferences. Do not add a separate linked-issues inventory. The generated PR record keeps source issues inline asRelated #NNNon the PR that shipped them - when backfilling an older linked-issues inventory, preserve reporter credit inline for every GitHub-confirmed closing PR relationship. Do not infer a PR relationship from a generic cross-reference event, invent an unrelated PR link for a standalone report, or recreate the retired inventory
- the complete contribution record lists every verified in-range PR and
explicitly retained seed-only PR exactly once as
**PR #NNN**. Discovery preserves canonical/cherry-pick provenance and requires frozen-history membership for contextual references; inline context alone cannot create a contribution row. It preserves author/co-author credit and any issue references in the original title - the provenance arithmetic and unique total must match the rendered PR rows exactly; candidate validation rejects malformed or forged counts
- direct commits remain in the manifest with GitHub-resolved author,
co-author, issue, and editorial-eligibility data. They inform grouped
prose but are never rendered as a public
#### Direct commitsdump. Add direct-commit credit to a grouped bullet only when it shares an explicit closing issue reference or at least two distinctive subject terms - the verifier rejects ordinary
docs,test,refactor,ci,build,chore, andstylePRs in Highlights, Changes, or Fixes. An explicit Conventional Commits!marker makes any type editorial-eligible; include its verified user-facing breaking change and migration guidance with the original PR ref and credit. Keep other internal contributions only in the complete PR record - classify conventional titles from their declared type and scope, not
incidental words such as
docorbuildin their descriptions. For untyped titles, retain internal-work signals such asQA,test,docs,refactor,lint, orCI; eligibility never replaces the source audit that establishes a user-visible outcome - do not add GHSA references, advisory IDs, or security advisory slugs to changelog entries or GitHub release-note text unless explicitly requested
- initial release generation keeps its existing credit policy: never thank
bots,
@claude,@codex,@openclaw,@clawsweeper, or@steipete. The separately approved post-docs GitHub body uses its complete verified human roster, including@steipetewhen credited - do not use GitHub's release contributor count as the source of truth; the changelog must carry the complete human credit set itself
- keep
-
Sorting preference:
- security/data-loss and content-boundary fixes
- transcript/replay/reply delivery correctness
- channels and mobile integrations
- providers/Codex/local model reliability
- install/update/release path reliability
- performance and observability
- docs and contributor-only/internal details last or omitted
-
Keep bullets single-line unless existing file style forces otherwise. Avoid internal release-process noise unless it changes user install/update safety.
-
Check release-note side conditions:
- inspect
src/plugins/compat/registry.ts - inspect
src/commands/doctor/shared/deprecation-compat.ts - if a deprecated compatibility record reaches
removeAfter, remove it when proven safe or move it toremoval-pendingand record the blocker; keep a dueremoval-pendingrecord only until its documented conditions are met
- inspect
-
Validate and ship:
- after the manifest-driven rewrite, regenerate and verify the complete
contribution record before committing:
node --import tsx .agents/skills/openclaw-changelog-update/scripts/verify-release-notes.mjs \ --base <base-tag> \ --target <target-ref> \ --main-ref origin/main \ --version <YYYY.M.PATCH> \ --manifest /tmp/openclaw-release-<YYYY.M.PATCH>.json \ --write-ledger - the command fails when any
#NNNreference in release history or the rendered release section cannot resolve, when reverted work is presented as shipped, when a source PR is absent from the contribution record, when direct commits are rendered as a public record dump, when non-editorial PRs appear in grouped prose, or when an eligible PR author or known co-author is missing from that PR'sThanks @...credit. It also fails before history collection when--baseis not an ancestor of--target, when### Highlightshas fewer than five or more than eight top-level bullets, or when the existing prose/record names a PR outside the source range. Only an explicit--seed-refmay add historical PR inventory; an explicit repeatable--shipped-refmay subtract PRs proven present in a prior shipped tag - when grouped prose names a PR, that same bullet must retain every contributor and linked-reporter credit from its generated PR record
- unqualified
#NNNreferences resolve againstopenclaw/openclaw; cross-repository references such asopenclaw/imsg#141remain literal text and must not be rewritten as local issue links - after the GitHub release or prerelease is published, verify every matching
release page against the same source section:
node --import tsx .agents/skills/openclaw-changelog-update/scripts/verify-release-notes.mjs \ --base <base-tag> \ --target <target-ref> \ --version <YYYY.M.PATCH> \ --release-tag v<YYYY.M.PATCH> \ --check-github - add one
--release-tagfor every beta and stable page in the train; a### Release verificationtail is permitted, but any other body drift fails the check scripts/render-github-release-notes.mtsis the canonical release-body renderer used by candidate validation, publish, and verification. When the complete## YYYY.M.PATCHsection fits GitHub's 125,000-character limit and the renderer's matching 125,000-byte safety ceiling, the body must contain that exact section including its heading- when the complete source section exceeds either limit, the renderer keeps the exact
grouped editorial notes through the line before
### Complete contribution record, then emits that heading with a stable link to the full contribution record in the tag-pinnedCHANGELOG/records/YYYY.M.PATCH.md(historical monolithic tags retain their originalCHANGELOG.mdrecord link). Never truncate a bullet or partial record, and never hand-author a different compact form - append
### Release verificationonly when it fits after the canonical full or compact body is chosen. If it does not fit, omit the body tail and retain the immutable attached release evidence; never compact a fitting full contribution record just to preserve the optional tail pnpm release:candidateperforms this deterministic render check from the exact target before it dispatches Full Release Validation, including when local generated checks are explicitly skippedgit diff --checkpnpm changelog:checkvalidates the split index, records, and marked docs mirrors- for docs/changelog-only changes, no broad tests are required
- stage the target release entry, its generated record, and changed index; commit with
git commit -m "docs(changelog): refresh YYYY.M.PATCH notes" - push the release branch without rebasing it onto moving
main - when all fixes and final notes are committed before fresh full qualification, record that commit as both Code SHA and Release SHA; use the same successful full parent/attempt and its exact publication bytes for both roles
- only when notes change after Code qualification, require
git diff --name-only <code-sha>..<release-sha>to includeCHANGELOG/YYYY.M.PATCH.mdand only that entry, its matching record, andCHANGELOG.mdbefore optionally usingsplit-changelog-release-v1. Additions or modifications of the selected entry/record are allowed; renames, deletions, other releases, and docs source changes are not. Historical root-only receipts retainchangelog-only-release-v1. The split path retains green Code proof and qualifies new Release SHA package bytes. Any other changed path requires fresh product qualification
Post-release docs mirrors
After explicit approval of the docs publication, publish the approved docs sources and their mechanically flattened changelog in the same source PR. This is separate from initial release generation; do not run an editorial rewrite automatically during release preparation or publication.
Render the ordered docs sources into one full Markdown file, even for a large release or a release spanning several docs pages:
pnpm changelog:from-docs --version YYYY.M.PATCH \
--source docs/releases/YYYY.M.PATCH.md \
--output CHANGELOG/YYYY.M.PATCH.md
pnpm changelog:check
Repeat --source in the approved reading order for a multipart release. The
converter removes presentation wrappers, turns accordion titles into headings,
expands docs links, and preserves source text, credits, code, tables, and images.
Unsupported markup fails instead of silently dropping content. Its first-line
marker binds the ordered source paths and exact source digest. Checks compare
only marked mirrors against their sources; untouched historical originals stay
in their initial format. Preserve CHANGELOG/records/YYYY.M.PATCH.md as the
frozen accounting record when replacing reader prose, and regenerate the mirror
in the same PR whenever its docs sources change.
The approved publication bundle owns source merge, verified docs deployment,
and the later GitHub Release body update as separate recorded steps. After
deployment, freshly read the existing release, then update only its body with
version/statistics, Raw changelog and docs links, one alphabetically deduplicated
verified human thanks roster, and the unchanged release-verification section.
Include verified PR, direct-commit, coauthor, and issue credit, including
@steipete; exclude bots. Keep both the 125,000-character and 125,000-byte
ceilings without truncating credits or proof. GitHub owns native avatars and
assets. Verify the body readback, resume only incomplete steps, and never retag,
rebuild, republish assets, or rerun initial publication to replace this body.
The initial history verifier rejects docs mirrors; use the docs-publication
workflow for their verification.
Extended-Stable Variant
Extended-stable has one release commit and no GitHub Release body. After version
prep and approved backports, regenerate ## YYYY.M.P with the regular manifest
and original-main-PR provenance rules. Land it by PR, then validate the final
branch tip before tagging. Re-audit after a product backport; a tooling-only
repair needs no changelog entry. Never rewrite a published tag or changelog.
Quota / API Outage Rule
If GitHub API quota is exhausted, do not idle. Continue work that does not need GitHub API:
- local changelog rewrite and release-note extraction
- local pretag checks and package/build sanity
- git push/tag checks over git protocol
- npm registry
npm viewchecks - exact workflow-dispatch command preparation
Only GitHub Release creation, workflow dispatch, run polling, artifact download, and issue/PR mutation need API quota.
Discovery context
Discovered by repository scan. No exact path reference found in the snapshot’s root AGENTS.md.