Files
holocron/plugins/git/skills/git-workflow/SKILL.md
Claude Code AI - Gitea MCP 598a7c326a refactor(skills): retrofit the corpus to the ADR-0020 context contract (#129)
Retrofits all 39 skills to ADR-0020's description/body context contract, then fixes what six rounds of independent review found in that retrofit — including four ways the hot gate itself failed open.

Closes #99, #107, #108, #110, #111, #114, #115, #120.

## The retrofit (waves 1-5)

| | Start | Now |
|---|---|---|
| Description FAILs (>400 chars) | 26 | **0** |
| Body FAILs (>900 words, body-only) | 9 | **0** |
| Dangling routing targets | 2 | **0** |
| `Kyberforge.CompositionNote` | 10 | **0** |
| Preload tax | 21,005 chars | **~10,500** |

Under the 12,000-char success criterion. Per-wave detail is on #99.

## The review fixes

**The gate failed open four ways, three of them found after the retrofit shipped.** An unrecognised follower token made a dangling target vanish. A skill directory with no `SKILL.md` resolved as a valid target, so a commit could be green locally and red in a fresh clone — three existing fixtures were relying on that, one of which made the install-leak A/B pass vacuously. Then the free-standing `/name` sweep turned out to be gated on the sentence carrying a boundary marker, so route notation in any other sentence was invisible — not an ERROR, not a SUGGESTION, not an INFO — which left the documented "`/name` always blocks" promise false from a second direction. All four fixed and pinned.

**Two checks were silently not running.** `validate-provenance.sh` checks 7-8 were dead across nine skills. Waking them exposed a deeper problem: they assume `Research doc:` names a source index, but 30 of 121 entries point at topic content documents, so every new check-7 INFO was a false positive and check 8 was saved from a false-FAIL flood only by an *unannounced* skip. Checks 7/8 are now scoped to source indexes and every skip announces itself (#121).

**The retrofit's own anti-goal, four times.** ADR-0020 warns that a blunt gate gets satisfied by deleting content rather than relocating it. `diagnose` and `skill-audit` relocated prose and then read it unconditionally; `prototype` and `vale-config` deleted rules outright that survived nowhere. All four addressed.

## Verification

- `bash tests/run-tests.sh --strict` — 24 suites, 0 skipped, 0 failed
- `bash tests/run-bats.sh` — 325 tests, 0 failures
- `pre-commit run --all-files` — 17/17
- `pre-commit run --hook-stage pre-push --all-files` — 16/16, with `apm marketplace check` and `apm pack --check-clean` run against the remote, not skipped
- `scripts/skill-size-check.sh` over all 39 skills — rc 0, 0 ERROR/FAIL, SUGGESTION-only
- Preload tax measured at **10,498 chars**, max description 390 — both inside budget
- Every new test proven non-vacuous by a deliberate mutation of the behaviour it covers

**Per-commit sync, stated accurately:** the ten commits from the latest review round each pass `check-plugin-content-sync` in isolation, verified by checking each out in a detached worktree with a clean between. The earlier gitea window (`dfacf05..bedbd1d`, nine commits) does **not** — its mirror was regenerated in one batch at `bbc7300`. An earlier revision of this description claimed the property held for every commit; it does not, and a bisect through that window lands on a red commit. **Squash-merge** to collapse it, or accept that this range is not bisectable.

## Version bump

Six plugins and the catalog take a **patch**, not a minor. The branch is **89 commits — 40 `fix` / 30 `refactor` / 12 `docs` / 5 `chore` / 2 `test` — zero `feat`, zero `!`, zero `BREAKING CHANGE`** — and adds no skill, agent, command or hook. (Two earlier revisions of this section cited a stale histogram, most recently 78 commits; the figures above are measured at HEAD.) Both rules this repo ships (`forge/references/version-bump.md`, landing in this PR, and `git-commits/references/conventional-commits-spec.md`) make that a patch, and the catalog set is unchanged at 7 entries.

Not settled by that: four published files were removed from the installed tree, three moved, and `caveman` gained `disable-model-invocation`, retiring its old triggers. Under a strict reading those are major-class and currently ship under `refactor:` with no marker. Whether the deployed skill surface is a public contract is written down nowhere — worth deciding, but it outlives this PR.

## Deliberately not in scope

#112 (cherry-pick ownership, now resolved in favour of `git-commits`), #113 (`rtk git` normalisation), #116 (research fan-out), #101 (audit-skill merge), #122 (non-spec skill-root files), #123 (no PRD producer) stay open. #117 is the one worth reading: the contract's remedy is to move prose into `references/`, which is exactly where neither the size gate nor Vale looks — and the blind spot is wider than #117 currently records, since there is no root `.vale.ini` at all, so every ADR, `CONTEXT.md` and `README.md` is unlinted too.

That blind spot let this branch carry two `level: error` `Kyberforge.SentenceOpenerThereIs` violations into `references/` files it created — `provider-adapter-author/references/provider-matrix.md:31` and `agent-audit/references/finding-criteria.md:95`. Both are reworded in `afadaae`, confirmed by routing each file through the audit's own `vale-wrap.sh` (1 error each before, 0 after). Five further occurrences sit in `references/` files already on `main`; those are the pre-existing corpus and stay with #117, which is the real fix.

Also unfixed and not this PR's: `apm install` appends a duplicate `SessionStart` entry to `.claude/settings.json`, so a fresh clone cannot get pre-push green without an edit AGENTS.md warns against. Reproduces identically on `main`.

Co-authored-by: Defame1297 <gitea@rkdr.net>
Reviewed-on: https://git.dev.rkdr.net/Defame1297/holocron/pulls/129
Co-authored-by: Claude Code AI - Gitea MCP <claude@noreply.git.dev.rkdr.net>
Co-committed-by: Claude Code AI - Gitea MCP <claude@noreply.git.dev.rkdr.net>
2026-09-01 13:47:46 +00:00

4.3 KiB

name, description, metadata
name description metadata
git-workflow Use when a human's local git request is general or ambiguous — it routes to the owning domain skill: `git-commits`, `git-branches`, `git-history`, `git-remotes`, `git-submodules` or `git-worktrees`. An unambiguous request goes straight to its domain skill instead. Not an agent caller -> `git-orchestrate`. Not Gitea -> `gitea-workflow`.
category source_keys
git
nvie-gitflow-post
atlassian-gitflow-tutorial
gitflow-cheatsheet
context7-git-htmldocs
org-git-conventions

Gotchas

  • Session context built during one multi-step request — branch names, the chosen base, the commit strategy — persists for that request and then clears. Do not re-ask the user for a decision they already gave you earlier in the same workflow.
  • Run parent-repo commands through rtk git <command>, never bare git <command>. This is a mandated org wrapper, not a style preference. Submodule-specific commands run from inside the submodule's own directory instead.

Domains

Route to the domain that owns the operation, and in sequence when a request spans two — a rebase that ends in a force-push is git-commits, then git-remotes. An unambiguous request should have gone straight to the domain skill; this one exists for the ones that did not.

The request is about Domain
Writing, amending, squashing, or cherry-picking a commit, and its message git-commits
Rebasing onto a new base, or undoing a commit with reset git-commits
Creating, switching, deleting, renaming, tracking, or merging a local branch git-branches
Stashing work in progress to move between branches git-branches
When a change landed, which commit broke something, what to revert or backport git-history
Anything touching a remote — remote config, fetch, push, pull — even unnamed git-remotes
A nested repository pinned inside this one by a recorded commit git-submodules
Several branches checked out at once, in separate directories, without stashing git-worktrees

git-orchestrate executes whatever this resolves to (step 5); the table only decides which domain owns the request.

Workflow

  1. Parse intent — extract the operation and, from the table above, the domain that owns it, plus any options the user named.
  2. Check the hard rules — if the request creates, amends, or rewrites a commit, pushes, or touches hooks, config, or credentials, read references/hard-rules.md. Raise the relevant rule before acting, not after.
  3. Read the repo — current branch, working-tree state, and which branching model the repo follows (the orchestrator reads branching_pattern from plugin config; infer from branch names if absent); the last of those decides which tips are worth offering.
  4. Gate destructive operations — before force-push, branch deletion, rebase, or force-checkout, show what will happen and ask "Proceed?". Cancel gracefully if the user declines. Never supply the confirmation on the user's behalf. Some operations are refusals, not confirmations: never offer "Proceed?" for a force-push of main or master.
  5. Invoke the git-orchestrate agent with operation, parameters (user-provided or inferred), context (step 3 plus the session context), and confirm: true only for a destructive op the user approved in step 4.
  6. Clarify when the orchestrator asks for more — put its question to the user in plain language ("Which branch should this be based on?") and loop back to step 5 with the answer.
  7. Report the outcome — on success, the result and what changed, in plain language; on failure, the error reason and a recovery action.

Interaction style

The caller is a human, so the interaction is the point. Explain each step and why it happens ("I'm squashing your last 3 commits into one clean commit" beats "Squashing commits"), show progress as you go, and prefer natural language to raw command lines.

Match tips to the repo's branching model rather than offering generic advice: on a Gitflow repo, feature branches come off develop and main tracks only released code; on a trunk-based or GitHub Flow repo, short-lived branches off main keep merges small and reviewable.