Add fragment-based changelog system to eliminate per-PR merge conflicts

[hujc7]

Problem

Every PR that touches source/<pkg>/ updates source/<pkg>/docs/CHANGELOG.rst and source/<pkg>/config/extension.toml. Both files are append-only at the top, so concurrent PRs collide on the same lines. Contributors burn time resolving merge conflicts that say nothing about the actual change. Maintainers occasionally land an entry under the wrong version heading because the resolution is mechanical and easy to mis-do.

Solution

Adopt the towncrier-style fragment-per-PR pattern (used by pip, urllib3, Twisted, Trio, attrs, Sphinx, …):

• Each PR adds one fragment file per touched package under source/<pkg>/changelog.d/<slug>.<tier>.rst. Different PRs touch different files — no conflict.
• A nightly CI workflow rolls accumulated fragments into per-package CHANGELOG.rst entries, bumps each extension.toml, deletes consumed fragments, and pushes the result back to develop.
• A PR gate rejects any PR that touches a managed package's source without adding a fragment, and rejects modifications to existing fragments (immutability).

CHANGELOG.rst and extension.toml become append-only-by-CI files; humans never edit them.

Contributor workflow (one fragment per touched package)

# Branch: jdoe/fix-spawn-pose
$EDITOR source/isaaclab/changelog.d/jdoe-fix-spawn-pose.rst

Fixed
^^^^^

* Fixed :meth:`~isaaclab.assets.Articulation.write_root_pose` ignoring the env-id mask.

Filename convention:

Filename	Effect
<slug>.rst	patch bump (X.Y.Z → X.Y.Z+1)
<slug>.minor.rst	minor bump (X.Y.Z → X.Y+1.0)
<slug>.major.rst	major bump (X+1.0.0) — breaking change
<slug>.skip	no entry, no bump (CI / docs / test-only PRs)

The slug is any short, unique name. Branch name with / replaced by - is the recommended default — already in scope at commit time, naturally unique because of the <user>/<feature> branch convention. Within a batch, the highest tier wins for the package (major > minor > patch).

What the nightly does

tools/changelog/cli.py compile --all walks every package's changelog.d/, sorts fragments by merge time, merges sections across fragments (canonical order: Added, Changed, Deprecated, Removed, Fixed), prepends a single X.Y.Z (YYYY-MM-DD) block to CHANGELOG.rst, bumps extension.toml, and deletes the consumed files. One commit per night, with a body listing every package that bumped:

[CI][Auto Version Bump] Compile changelog fragments (schedule)

Bumped packages:
- isaaclab: 4.6.26 → 4.7.0
- isaaclab_physx: 0.5.29 → 1.0.0
- isaaclab_tasks: 1.5.32 → 1.5.33

What's tested

Unit + integration (82 tests, all passing): filename regex acceptance/rejection, bump-tier aggregation, cross-fragment section merge, immutability, missing-fragment-per-touched-package, version-pinning guards, content validation. Worked-example fixtures under tools/changelog/tests/fixtures/integration/{01_patch_bump, 02_minor_bump, 03_major_bump}/ double as living demos.

Live end-to-end on a fork (real GitHub Actions runners, not local mocks):

1. Single fragment dry-run + real run: ✓ produces correct CHANGELOG block, bumps extension.toml, deletes fragment, commits with the expected subject/body.
2. Multi-package multi-tier batch (7 simulated PRs across 3 packages, mixed Fixed/Added/Deprecated/Removed/Changed/Breaking, mixed patch/minor/major/skip): ✓ each package bumps to the right tier (isaaclab: 4.6.26→4.7.0 minor, isaaclab_physx: 0.5.29→1.0.0 major, isaaclab_tasks: 1.5.32→1.5.33 patch), single commit lists all three.
3. Cross-fragment section merge (3 fragments → one Added section with 3 bullets, 2 fragments → one Fixed section with 2 bullets, one fragment contributing to both): ✓ bullets correctly merged in merge-time order.

Fork test branch: https://github.com/hujc7/IsaacLab/tree/test-changelog-nightly.

Setup checklist for upstream after merge

The new workflow files are pinned by SHA per the existing CI conventions; no additional repo settings are required for the workflows to run. Two optional follow-ups maintainers may want:

1. Add CHANGELOG_PAT secret (fine-grained PAT with contents:write on the repo). Optional. Without it, the nightly's auto-commit uses GITHUB_TOKEN — sufficient to push, but pushes signed with GITHUB_TOKEN do not trigger downstream workflow runs (this is GitHub's by-design loop guard). Adding the PAT lets the auto-commit re-trigger the docs / Docker rebuilds.
2. Branch protection on develop may need an exception for github-actions[bot] so the auto-commit can push directly. If develop already accepts non-PR pushes from CI service accounts, no change.

Files changed

• tools/changelog/cli.py — single entry with compile and check subcommands.
• tools/changelog/tests/ — 82 tests + integration fixtures.
• .github/workflows/changelog-check.yml — PR gate (per-PR).
• .github/workflows/nightly-changelog.yml — nightly auto-compile (cron + workflow_dispatch).
• AGENTS.md — contributor docs (+10 / -5 lines in the existing "Changelog" section).
• source/<pkg>/changelog.d/.gitkeep — empty directory placeholders for every managed package.

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

This PR introduces a fragment-based changelog system to eliminate merge conflicts on CHANGELOG.rst and extension.toml. Contributors drop per-PR fragment files that are compiled nightly into the canonical changelog. The implementation is well-designed and solves a real pain point, but has several bugs in the CI workflow configurations and a logic error in the checker script that could cause false negatives.

Architecture Impact

Self-contained tooling addition. No impact on runtime code paths. The MANAGED_PACKAGES list is duplicated across two files (changelog.py and check_changelog.py) — not ideal but acceptable for standalone scripts. The nightly workflow commits directly to the branch, which is standard practice but worth noting for teams with protected branches.

Implementation Verdict

Minor fixes needed — the workflow files have action pinning violations and a variable reference bug that will cause the changelog check to fail in PR contexts.

Test Coverage

No tests provided. For tooling scripts this is acceptable if manually verified, but given the complexity of fragment parsing, version bumping, and RST generation, unit tests for parse_fragment(), merge_fragments(), bump_version(), and nightly_version() would significantly reduce regression risk during future maintenance.

CI Status

No CI checks available yet.

Findings

🔴 Critical: .github/workflows/changelog-check.yml:42 — Wrong variable used for base ref in PR context

The workflow uses ${{ github.event.inputs.base_ref }} which is only populated for workflow_dispatch. When this workflow is later enabled with a pull_request trigger, this variable will be empty, causing the check to always pass (git diff against empty string may behave unexpectedly or error).

- name: Verify changelog fragments
  run: python3 tools/check_changelog.py ${{ github.event.inputs.base_ref }}

Should be:

run: python3 tools/check_changelog.py ${{ github.event.inputs.base_ref || github.base_ref }}

Additionally, line 39 uses ${{ github.base_ref }} for the fetch but then line 42 uses the input variable — inconsistent and will break when the trigger changes.

---

🔴 Critical: .github/workflows/changelog-check.yml:30,35 and nightly-changelog.yml:27,34 — Actions not pinned by SHA hash

Per AGENTS.md: "Pin actions by SHA hash. Use action@<sha> # vX.Y.Z format for supply-chain security."

- uses: actions/checkout@v6
- uses: actions/setup-python@v5

These must be pinned to SHA hashes per project policy.

---

🟡 Warning: tools/changelog.py:287-289 — Fragment count in log message is misleading when some fragments are empty

print(f"  {package_dir.name}: {len(fragments_paths)} fragment(s) → version {new_version}")

This prints the count before filtering out empty fragments, so if 3 fragments exist but 1 is empty, it reports "3 fragment(s)" but only 2 are actually compiled. Should use len(parsed) after the filtering on line 283.

---

🟡 Warning: tools/check_changelog.py:77-82 — False negative when fragment added outside the PR's changed files

The logic checks if a fragment file is in changed (files changed in this PR). However, if someone adds a fragment in a separate commit that's already on the base branch, or if a fragment was pre-created, the check would fail even though a valid fragment exists in the directory.

Consider also checking if any .rst or .skip file exists in the changelog.d/ directory on disk, not just in the diff. This is a design choice, but the current behavior may surprise contributors.

---

🟡 Warning: tools/changelog.py:246 — _insert_after_header regex assumes specific header format

m = re.search(r"^-+\n\n", text, re.MULTILINE)

This requires exactly two newlines after the dashes. If an existing CHANGELOG.rst has different spacing (e.g., one newline, or trailing whitespace on the blank line), the insertion will fail with a confusing error. Consider a more lenient pattern like r"^-+\s*\n\s*\n".

---

🔵 Improvement: tools/changelog.py and tools/check_changelog.py — Duplicated MANAGED_PACKAGES list

Both files define identical MANAGED_PACKAGES lists (lines 55-66 in changelog.py, lines 27-38 in check_changelog.py). If a package is added/removed, both must be updated. Consider extracting to a shared module or a config file, though for standalone scripts this duplication is tolerable.

---

🔵 Improvement: .github/workflows/nightly-changelog.yml:44-46 — git add pattern may stage unintended files

git add source/*/changelog.d/ \
        source/*/docs/CHANGELOG.rst \
        source/*/config/extension.toml

The glob source/*/changelog.d/ will match any subdirectory, potentially including unmanaged packages. This is probably fine since only managed packages will have fragment deletions, but explicitly listing the managed packages would be safer and self-documenting.

---

🔵 Improvement: tools/changelog.py:173-174 — Version bump doesn't validate explicit --version argument

When using --version X.Y.Z, there's no validation that the string is a valid semver or PEP 440 version. Invalid input like --version "foo" will be written directly to extension.toml, potentially breaking downstream tooling.

[greptile-apps]

Greptile Summary

This PR introduces a towncrier-style fragment-based changelog system to eliminate per-PR merge conflicts on CHANGELOG.rst and extension.toml. Each PR deposits a small changelog.d/<pr>[.minor|.major].rst file; a nightly workflow (or manual invocation) compiles accumulated fragments into versioned entries and deletes the consumed files.

• P1 — nightly workflow targets the wrong branch: nightly-changelog.yml has no ref: develop in its actions/checkout step. GitHub's schedule trigger checks out the repository's default branch; if that is not develop, the nightly will compile an empty fragment set on the wrong branch and push nothing useful back to develop. Add ref: develop to the checkout and git push origin HEAD:develop to the push step.

Confidence Score: 4/5

Safe to merge after fixing the nightly workflow checkout branch — core logic, PR gate, and tests are solid.

One P1 in the nightly workflow (missing ref: develop) caps the score at 4. The CLI itself is well-tested (75 tests, integration tests against worked examples), and the PR gate workflow correctly handles both pull_request and workflow_dispatch triggers. The P2 dry-run UX issue in cli.py does not affect correctness.

.github/workflows/nightly-changelog.yml — missing explicit branch target in checkout and push.

Important Files Changed

Filename	Overview
tools/changelog/cli.py	Core changelog compiler and PR gate — 968 lines; logic is well-structured with clean class hierarchy. Minor UX gap in dry-run output when only skip files are present.
.github/workflows/nightly-changelog.yml	Nightly cron workflow — missing ref: develop in checkout means scheduled runs operate on the default branch instead of develop; git push without refspec compounds the issue.
.github/workflows/changelog-check.yml	PR gate workflow — correctly resolves base ref for both pull_request and workflow_dispatch triggers using a dedicated 'Resolve base ref' step.
tools/changelog/tests/cases/test_validate.py	Validation tests — comprehensive coverage of filename patterns, empty files, missing section headings, bullet-free sections, immutability, chain tolerance, and argparse guards.
tools/changelog/tests/cases/test_integration.py	End-to-end integration tests — drives the compiler against all three worked examples and byte-compares output; keeps examples as living documentation.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    PR[Pull Request] -->|adds fragment file| CD["source/<pkg>/changelog.d/<pr>[.minor|.major].rst"]
    PR -->|triggers| Gate["changelog-check.yml\n(PR gate)"]
    Gate -->|git diff origin/base...HEAD| Diff[PRDiff.from_git]
    Diff -->|evaluate rules| Rules{"3 rules:\n1. Immutability\n2. Content validity\n3. Owns fragment"}
    Rules -->|pass| GateOK[✓ check passes]
    Rules -->|fail| GateFail[✗ exit 1]
    Cron["nightly-changelog.yml\n(0 5 * * * UTC)"] -->|checkout default branch ⚠️| Compile["cli.py compile --all"]
    Compile --> Batch["FragmentBatch.from_dir\nper package"]
    Batch -->|sort by merge_time| Sorted[Sorted fragments]
    Sorted -->|aggregate_bump| Bump["max bump tier\n(major > minor > patch)"]
    Bump -->|compile_to_entry| Entry[RST entry block]
    Entry -->|write_changelog_entry| CL[CHANGELOG.rst updated]
    Entry -->|write_version| TOM[extension.toml bumped]
    CL & TOM -->|delete_all| Clean[Fragments deleted]
    Clean -->|git add + commit + push| Develop[develop branch]

Loading

_{Reviews (2): Last reviewed commit: "Restore nightly auto-compile workflow" | Re-trigger Greptile}

[greptile-apps]

github.base_ref is empty for workflow_dispatch triggers

The Fetch base branch step uses ${{ github.base_ref }}, which is only populated for pull_request events. When this workflow is triggered via workflow_dispatch, github.base_ref is an empty string, so the step effectively runs git fetch origin with no branch argument — either fetching all branches unexpectedly or silently failing. The Verify changelog fragments step immediately after correctly uses ${{ github.event.inputs.base_ref }}, but the fetch using the wrong context means origin/<base_ref> won't exist when the checker runs git diff --name-only origin/<base_ref>...HEAD.

Suggested change

	- name: Fetch base branch
	run: git fetch origin ${{ github.event.inputs.base_ref }}

[greptile-apps]

.skip files accumulate indefinitely after nightly runs

compile_package cleans only *.rst fragments (fragment_dir.glob("*.rst")), but .skip opt-out files are never removed. After the nightly job runs and deletes the .rst fragments, any .skip files remain in changelog.d/ permanently. Since subsequent nightly runs find no .rst files they return early, so .skip files accumulate in the repository with every PR that opts out.

To fix, also glob and remove .skip files under --clean:

if clean and not dry_run:
    for p in fragments_paths:
        p.unlink()
    for p in fragment_dir.glob("*.skip"):
        p.unlink()
    print(f"  {package_dir.name}: deleted {len(fragments_paths)} fragment(s).")

[greptile-apps]

Fragment count logged before empty fragments are filtered

Line 283 logs len(fragments_paths) as the number of compiled fragments, but fragments_paths is not updated after empty-parsed fragments are silently dropped (line 274: parsed = [s for s in parsed if s]). If one or more fragments parse as empty, the printed count will be higher than the number actually merged into the changelog, giving a misleading message to operators.

Suggested change

	print(f" {package_dir.name}: {len(fragments_paths)} fragment(s) → version {new_version}")
	print(f" {package_dir.name}: {len(parsed)} fragment(s) → version {new_version}")

[greptile-apps]

_insert_after_header regex matches first dash-underline in the file

The regex r"^-+\n\n" finds the first line of dashes followed by a blank line. This is designed to match the top-level Changelog\n---------\n title underline. However, if any existing version entry content (e.g., a code block, note, or admonition) contains a dashes-only line followed by a blank line before the first version heading, the insertion point will be wrong and new entries will land in the middle of the existing changelog. Using a more specific anchor — e.g., matching Changelog\n followed by the dashes — would be more robust.

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

The new commits addressed the critical action pinning violations and the base ref variable issue. The workflow files now correctly use SHA-pinned actions and properly resolve the base ref for both workflow_dispatch and pull_request triggers.

Previous Concerns Status

• ✅ Action SHA pinning — Addressed. Both workflows now use @de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6 and @a26af69be951a213d495a4c3e4e4022e16d87065 # v5.
• ✅ Base ref variable bug — Addressed. Line 40 now uses ${{ github.event.inputs.base_ref || github.base_ref }} and line 43 correctly references ${{ steps.base.outputs.ref }}.
• ✅ Header regex leniency — Addressed. The regex on line 246 is now r"^Changelog\n-+\s*\n\s*\n" which handles whitespace variations.

CI Status

• ❌ pre-commit: failure — Needs investigation; may be unrelated to this PR's changes.
• ❌ Build cuRobo Docker Image: failure — Appears unrelated to changelog tooling.

Implementation Verdict

Ship it — the critical issues have been resolved. The remaining warnings from the previous review (fragment count logging, false negative edge case, duplicated MANAGED_PACKAGES) are minor and acceptable for initial merge.

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

No new commits since the previous review — the HEAD remains at f8a11d33. All previously identified critical issues (action SHA pinning, base ref variable bug, header regex leniency) were addressed in the earlier commits.

Implementation Verdict

Ship it — the implementation is solid and ready for merge. The CI failures (pre-commit, Build cuRobo Docker Image) appear unrelated to the changelog tooling changes.

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

The new commit d630bc9d updates the GitHub Actions SHA pins from v4/v5 tags to the correct current versions (actions/checkout to v6 SHA, actions/setup-python to v5 SHA). This addresses the previous concern about action pinning. No new issues introduced.

Implementation Verdict

Ship it — the implementation is solid and the CI checks are passing. The action SHA pins now follow the repo's security guidelines.

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

The new commit 5956248f adds a missing AGENTS.md changelog fragment file for this PR (source/isaaclab/changelog.d/5434.skip). This is consistent with the PR's own fragment-based system — since this PR only adds tooling/documentation and doesn't modify the isaaclab package's runtime code, a .skip file is the correct choice.

Implementation Verdict

Ship it — previous concerns about action SHA pins were addressed, and this final commit correctly adds the required changelog fragment opt-out for the tooling-only PR.

[hujc7]

@greptileai Review

[greptile-apps]

Nightly runs on the default branch, not develop

The actions/checkout step has no ref: field, so GitHub's schedule trigger checks out the repository's default branch (typically main). The PR description explicitly says the nightly "pushes the result to develop", but the workflow silently operates on whatever the default branch is. If the default branch is not develop, the nightly will compile main's (empty) fragment dirs, find nothing, and the real accumulated fragments on develop are never compiled. The git push without a refspec would also push to origin/<default_branch>, not origin/develop.

Add an explicit ref to the checkout and a target branch to the push:

Suggested change

	# to GITHUB_TOKEN which is sufficient for the push itself.
	token: ${{ secrets.CHANGELOG_PAT || secrets.GITHUB_TOKEN }}
	# Full history so the compiler can resolve each fragment's merge
	# time via ``git log --diff-filter=A --first-parent``.
	fetch-depth: 0
	- uses: actions/setup-python@a26af69be951a213d495a4c3e4e4022e16d87065 # v5
	- uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6
	with:
	ref: develop
	token: ${{ secrets.CHANGELOG_PAT || secrets.GITHUB_TOKEN }}
	fetch-depth: 0

And in the "Commit and push" step:

git push origin HEAD:develop

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

The new commit 17d1c61d removes the source/isaaclab/changelog.d/5434.skip file that was added in the previous commit. This is incorrect — the PR still modifies files under source/isaaclab/ (the changelog.d/.gitkeep file), so the CI changelog check will now fail for this PR.

Implementation Verdict

Minor fix needed — the skip file removal breaks the PR's own CI gate.

Findings

🔴 Critical: Missing changelog fragment for isaaclab package

The previous commit correctly added source/isaaclab/changelog.d/5434.skip to satisfy the changelog check, but this commit (17d1c61d) removes it. The PR adds .gitkeep files under source/isaaclab/changelog.d/ (and other packages), which counts as touching those packages. Without the skip file, the changelog-check.yml workflow will fail with "Missing changelog fragments for the following packages: isaaclab".

Fix: Restore source/isaaclab/changelog.d/5434.skip (and similar skip files for the other 10 packages that have new changelog.d/.gitkeep files added).

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

This PR introduces a comprehensive towncrier-style fragment-based changelog system to eliminate merge conflicts on CHANGELOG.rst files. The implementation is well-architected with thorough test coverage and clear documentation.

Architecture Impact

Self-contained tooling addition. The new tools/changelog/cli.py is invoked by two new GitHub workflows (changelog-check.yml for PR gates, nightly-changelog.yml for auto-compilation). No existing code paths are modified beyond AGENTS.md documentation updates. The empty .gitkeep files in each package's new changelog.d/ directory establish the fragment directories without triggering the changelog requirement (since they're not under source/<pkg>/ proper source paths).

Implementation Verdict

Ship it

Test Coverage

Thorough. The test suite covers:

• Fragment parsing and validation (test_parse.py, test_validate.py)
• Bump tier inference and aggregation (test_bump_suffix.py)
• RST formatting and version math (test_format.py)
• End-to-end compilation against worked examples (test_integration.py)
• PR gate rules including chained-PR tolerance, immutability, and skip files

The integration tests use the examples/changelog/ directories as living documentation that can't drift from actual behavior.

CI Status

The changelog-check workflow shows ✅ success, indicating the PR correctly handles its own fragment requirement. The .gitkeep files under source/<pkg>/changelog.d/ don't trigger the "must add fragment" rule because the check only fires when source files outside changelog.d/ are modified.

Findings

Previous concern about the missing skip file is not applicable — the CI is passing because adding .gitkeep files inside changelog.d/ directories doesn't count as "touching the package's source" (the check explicitly excludes changelog.d/ from the source-change detection). The design is correct.

No new issues found. The implementation is clean and well-tested.

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

This is a follow-up review on the same commit (c69876b). The previous review found no issues and recommended shipping. No new commits have been added since that review.

Implementation Verdict

Ship it

Findings

No new changes to review. The previous assessment stands — the implementation is well-architected, thoroughly tested, and the CI is passing. The fragment-based changelog system is ready to merge.

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

This is a follow-up review on commit cde8480. The changes since the previous review (c69876b) appear to be minimal — the diff and file contents are identical to what was previously reviewed. The implementation remains well-architected with comprehensive test coverage.

Implementation Verdict

Ship it

CI Status

Most checks are still pending, but core checks (pre-commit, Build Wheel, Check changelog fragments) have passed. No failures detected.

Findings

No new issues. The previous assessment stands — the fragment-based changelog system is ready to merge.

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

Follow-up review on commit f56ece5. The changes since the previous review (cde8480) appear to be cosmetic or non-existent — the implementation remains identical to what was previously approved.

Implementation Verdict

Ship it

CI Status

Core checks passing (pre-commit ✅, Build Wheel ✅, Check changelog fragments ✅). The Check for Broken Links failure appears unrelated to this PR's changes (likely a pre-existing flaky check or external link issue).

Findings

No new issues. The implementation is solid, well-tested, and ready to merge.

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

Follow-up review on commit ab9535f. The diff between the previous review (f56ece5) and this commit shows no changes to the implementation — the files are identical.

Implementation Verdict

Ship it

CI Status

Core checks passing (pre-commit ✅, Build Wheel ✅, Check changelog fragments ✅). The Check for Broken Links failure is unrelated to this PR.

Findings

No new issues. Ready to merge.

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

Follow-up review on commit c6d45db. The diff between the previous review (ab9535f) and this commit shows no substantive changes to the implementation — the files remain identical.

Implementation Verdict

Ship it

CI Status

Core checks passing (pre-commit ✅, Build Wheel ✅, Check changelog fragments ✅). The Check for Broken Links failure is unrelated to this PR (appears to be a pre-existing infrastructure issue).

Findings

No new issues. The fragment-based changelog system is well-designed with comprehensive test coverage. Ready to merge.

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

Follow-up review on commit 34637ce. The diff between the previous review (c6d45db) and this commit shows no changes to any implementation files — the PR remains identical.

Implementation Verdict

Ship it

Findings

No new issues. The previous review already approved this well-designed fragment-based changelog system. Ready to merge.

[isaaclab-review-bot]

🤖 Isaac Lab Review Bot

Summary

Follow-up review on commit 2f7ffbf. The diff between the previous review and this commit shows no implementation changes — the PR remains identical to what was previously approved.

Implementation Verdict

Ship it

CI Status

The "Check for Broken Links" failure appears unrelated to this PR's changes (changelog tooling doesn't affect documentation links). All relevant checks (pre-commit, changelog fragment check, wheel build) pass.

Findings

No new issues. Previous approval stands — this is a well-designed, thoroughly tested fragment-based changelog system ready to merge.