fix(credit): preserve v0.9.1 integration authorship

[Hmbown]

What changed

• map Hunter Bown and Fred Leitz to canonical numeric GitHub noreply identities
• resolve only the exact existing public-surface and PR fix(shell): default no-cwd shell commands to context.workspace #4673 commit identities through AUTHOR_MAP
• retain the original commits and merge ancestry without rewriting
• add negative tests proving changed SHAs or identities still fail

Validation

• cargo fmt --all -- --check
• python3 -m unittest scripts/test_check_coauthor_trailers.py
• python3 scripts/check-coauthor-trailers.py --range origin/main..HEAD --check-authors
• git diff --check origin/main...HEAD

This is a narrow prerequisite for direct contributor integration and PR #4675. It does not deploy, publish, tag, or release.

[cursor]

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[claude]

Claude finished @Hmbown's task in 3m 44s —— View job

---

Review: preserve v0.9.1 integration authorship

• Gather context (PR diff, custom instructions)
• Understand the request (code review)
• Review .github/AUTHOR_MAP changes
• Review scripts/check-coauthor-trailers.py changes
• Review scripts/test_check_coauthor_trailers.py changes
• Post final review

Traced the new PRESERVED_MAPPED_IDENTITY_EXCEPTIONS gate by hand against all three added unit tests (line-by-line through validate()); the logic checks out for the cases the tests cover. No provider/model/route concerns (Python credit tooling, unrelated to the v0.8.65 EPIC #2608 invariant) and no security issues (no shell injection, subprocess still uses list args, no secret handling changed).

1. SHA authenticity is untested and I could not independently verify it (worth confirming before merge)
scripts/check-coauthor-trailers.py:60-85 hardcodes three commit SHAs (5087269..., e37df06..., 6d0ebc88...) that the exception logic trusts unconditionally. The new tests in scripts/test_check_coauthor_trailers.py:109-168 only exercise these SHAs against synthetic mod.Commit(...) objects built to match the claimed author/co-author/subject — they never check the SHAs against real git history, so nothing in this PR proves these three commits actually exist on origin/main with the exact name/email/subject encoded here. Since credit-preservation is the entire purpose of the PR, a typo'd or mismatched SHA would make an entry either silently inert or (less likely, given the exact-tuple match) misapplied to the wrong commit. I tried to confirm the SHAs myself via git cat-file -t <sha> but this checkout is shallow (only d9fdee8 is present as the tip of origin/main) and outbound git fetch --unshallow was blocked in this environment, so I couldn't complete the check. Worth the author re-confirming against the real v0.9.1 integration history (e.g. git log <sha> -1 --format='%an <%ae>%n%(trailers:key=Co-authored-by)') before merge.

2. Minor: repeated lookups in the coauthor loop (non-blocking)
scripts/check-coauthor-trailers.py:262-281 calls lookup_identity(aliases, coauthor.email, coauthor.name) and is_preserved_mapped_identity(commit, "coauthor", coauthor) once in the error-reporting loop and again in the coauthor_emails-building loop for the same coauthor. Harmless (coauthor lists per commit are tiny) but could be collapsed into one pass if you want to avoid the duplication.

Good choices:

• The exception key is a strict 4-tuple (sha, role, normalized name, normalized email) — matches the existing LEGACY_AUTOMATION_TRAILER_EXCEPTIONS pattern in the same file, and the negative tests (test_preserved_hunter_credit_resolves_only_for_exact_commit, test_preserved_fleitz_credit_resolves_only_for_exact_commit) correctly prove a changed SHA or email still hard-fails, so the gate can't silently widen over time.
• AUTHOR_MAP additions (.github/AUTHOR_MAP:12-15) follow the existing format and resolve to canonical numeric noreply addresses; load_author_map's own validation (exercised via setUpClass in the test file) already guards against malformed entries.
• Docstring update at the top of the script clearly documents the new exact-SHA exception carve-out.

Note: I wasn't able to execute cargo fmt, the unittest suite, or check-coauthor-trailers.py directly in this sandbox (script execution was blocked pending approval), so I relied on manual trace-through of validate() against the new tests rather than running them. If you'd like me to actually execute the validation commands from the PR description, the --allowedTools for this job would need to permit running python3 against files under scripts/.

[Hmbown]

Verified against the fetched v0.9.1 integration/contributor objects before merge:

• 5087269606fc8847487b0a8b51ef6adffa8eb2ca: author Hunter B <hmbown@gmail.com>; subject docs(public): refresh v0.9.1 product truth; co-author Hunter Bown <hmbown@gmail.com>.
• e37df06caeb3064b2bb9263c1c98a903738f3a0a: author Hunter B <hmbown@gmail.com>; subject docs(web): make the homepage product first; co-author Hunter Bown <hmbown@gmail.com>.
• 6d0ebc881a8bd2469c45b25f2a606fa63681e112: author Fred Leitz <fred.leitz@gmail.com>; subject fix(shell): default no-cwd shell commands to context.workspace; co-author fleitz <fleitzo@gmail.com>.

Local validation also passed python3 -m unittest scripts/test_check_coauthor_trailers.py, python3 scripts/check-coauthor-trailers.py --range origin/main..HEAD --check-authors, cargo fmt --all -- --check, and git diff --check. The exact-SHA exceptions are intentionally narrow so the original contributor commits remain byte-for-byte intact.