Repository navigation
Description
Problem
Native commands do not complete the migration until both generated harness workflows invoke them. The current entry instructions at packages/agents/source/skills/rebase/SKILL.codex.md:16 and packages/agents/source/skills/resolve-rebase-conflicts/SKILL.codex.md:24 still invoke shell. Activation must use the launcher from #321 and retain the frozen fallback through N-2 without introducing another resolver.
Acceptance criteria
- AC1: Given a supported installed binary, when either generated skill workflow is followed, then it dispatches to the corresponding Go operation through the shared launcher.
- AC2: Given the same supported fixture, when a generated Go workflow runs, then its status, streams and repository effects match the frozen workflow except for recorded migration differences.
- AC3: Given a legacy invocation during the N-2 window, when its frozen fallback is selected through the established compatibility contract, then its prior behavior remains available.
- AC4: Given the activation diff, when the retained fallback is compared with the baseline, then its executable logic is unchanged.
- AC5: Given missing or too-old binaries, when the Go entry is invoked, then the outcome is exactly the shared launcher's documented version contract.
- AC6: Given either harness package is generated and deployed, when generated consistency is checked, then source, packages, token metadata and lock records agree.
- AC7: Given entry invocation from a repository subdirectory, when either workflow runs, then it resolves the same repository and prints usable follow-up guidance.
- AC8: Given the published migration guidance, when it is read, then the first required release and at least two warning releases are identifiable before fallback deletion.
- AC9: Non-functional: activation introduces no shell selection logic or second binary resolver.
- AC10: Non-functional: native workflow cases execute without skips on macOS, Linux and Windows.
Scope and non-goals
Scope: skill manifests, shared prose, command routing integration and end-to-end compatibility evidence for rebase and conflict resolution.
Non-goal: implementing or modifying the shared launcher (#321), moving or deleting frozen implementations, changing PreToolUse hooks (#328 and #753), or retiring Bats/Python coverage. Existing public entry paths stay available throughout N-2. Python/uv removal applies to native Go commands only; the frozen compatibility path retains its prerequisites.
Approach
Consume #321's launcher and version/fallback contract; identify its exact invocation before editing the skill templates. This child changes generated workflow instructions and integration coverage, not shell algorithms. Keep frozen source in place; any routing or retention work needed by the launcher is a prerequisite delivered by #321, not an uncounted copy or relocation here. Retain current entry conventions in compatibility checks. Generate from canonical templates following tools/generate-skills.sh:30 and verify both harnesses. Keep hook commands at packages/agents/source/skills/rebase/SKILL.claude.md:7 and packages/agents/source/skills/resolve-rebase-conflicts/SKILL.claude.md:7 unchanged. Document intentional loss of Python/uv prerequisites on the Go path and any launcher-selected missing-binary behavior separately from behavioral parity.
Dependencies
Blocked by: #895, #896, #897, #898, #899, #900, #901, #902, #903, #851 (the readiness delivery), #333 (verified binary assets), #544 (consumer installation), #753 (protection).
Prerequisite: #321's single launcher and frozen-fallback contract. Use delivered outcomes rather than requiring whole-area closure. The command release must contain all required operations before deployment. PreToolUse replacement remains with #328 under #753; activation must not weaken the existing guards.
Size
M. Forecast 600 code lines: 450 integration tests/fixtures and 150 Go routing/diagnostic integration. No frozen-source relocation is budgeted or authorized here. Prose allowance: 600 lines including maintained templates and migration guidance.
scc could not run in the isolated checkout. Measured baselines count non-blank lines other than full-line # comments, including Python docstrings and embedded code conservatively. Forecasts include implementation, tests, fixtures and registration; they are estimates, not measured future Go. Generated output is excluded only under packages/agents/claude/.apm/skills/, packages/agents/codex/.apm/skills/, .claude/skills/, .agents/skills/ and tokens.json. The source under packages/agents/source/ and copies anywhere else count. The final diff must be at most 800 code lines and 2,000 prose lines, with no exception; split before implementation if either forecast grows beyond its limit.
Verify plan
- AC1, AC2, AC7: run both generated workflows against isolated repositories, including a relative invocation from a subdirectory; compare streams, statuses and effects with the frozen implementations.
- AC3, AC4: record fallback hashes and run supported legacy cases under the Port the shared skill library (lib.sh, agent.sh, signing.sh, gates.py) to Go #321 compatibility contract; compare executable logic byte-for-byte.
- AC5, AC9: exercise absent, too-old and supported binaries through the one shared launcher; inspect the diff for duplicate resolvers or new shell branches.
- AC6: regenerate skills, refresh tokens/deployments and lock records, then run generated-consistency and packaging checks.
- AC8: compare migration documentation with the actual command-bearing release and the warning-release evidence.
- AC10: inspect native three-platform Go CI; keep shell-required integration evidence explicitly marked by executing platform.
- Definition of done: D1 each AC is verified at the merged head, with the commands and results cited in the pull request's Verification. D2 an independent review approves the exact merged head, and every finding has a recorded disposition. D3 required checks pass with no gate loosened. D4 the leak scans the repository specifies are clean, and every post made for this work passes the per-post leak review before it is posted. D5 every commit is signed. D6 the documentation matches the change or the Verification says why there is no impact. D7 every commit has an
Assisted-bytrailer and the pull request has theprovenance:agent-assistedlabel. D8 the pull request stays within the size limits. D9 follow-ups are filed as issues and listed under Related. D10 the merge closes the issue through a closing reference, the board shows Done, and the finalStatus:comment names the merge commit.
Risks
Activating before binaries or compatibility routing are available strands consumers. A blanket rename could silently exceed the size limit. Undo: revert the workflow activation and regenerate; keep the unchanged fallback and recovery evidence available.
Operator authority
None.
Metadata
Metadata
Assignees
Labels
Type
Fields
Priority
Work type
Projects
- StatusShow more project fieldsBacklog
Activity
tbhb-agent commentedon Oct 27, 2026
Refinement: READY — Activate both skill workflows through the shared launcher
Model: gpt-6-astra
Child 10 Activate both skill workflows through the shared launcher
Verdict: READY
Checked: Activation still requires command parity, frozen-fallback preservation and prerequisite delivery. The corrected proof and classification contracts fit its existing integration scope. Forecast remains 600 lines.
Gaps: None.
Split: None required.
Definition of done
The definition of done for this issue (Feature). Tick each item and give its evidence.
check-leaks.shwith the repository's private-strings list, and the leak review of each post.docs/, CHANGELOG or help text), or the Verification section says "No doc impact" and why.provenance:agent-assistedlabel.Status:comment names the merge commit.mise run checkis green, and so are the coverage floorsmise run cover-go-checkandmise run cover-py-check, whichcheckdoesn't run.packages/agents/source/commits whatmise run generate-skills, thenmise run skill-tokens, thenapm installrewrite, in that order, andcheck-generated-skills(part oflint) passes.docs/agent-skills-migration.mdor the adoption docs, or the PR's Verification says "No consumer impact" and why.Round: 2 of 3