sync validate function with commit name pasring to identify naming issues early

[timwelapt]

I needed a re-naming loop to pass rivet-commit. suggestion is to sync validate function so such an naming issue gets caught early. BR, Tim.

Claude-Opus-48 told:

Found the root cause. In rivet's commit-trailer grammar (rivet-core/src/commits.rs), a valid artifact ID is is_artifact_id: the prefix segments must be all uppercase letters — no digits. So UDS-102, TRACE-104, BUILD-100 pass, but MAD1-101 is rejected (MAD1 has a digit) and flagged as a malformed ref. rivet's YAML validator is more lenient than its commit parser, which is why these IDs were valid as artifacts but not as trailer references.

[avrabe]

Triage from the issue-fixing routine (first pass — welcome, Tim).

The root-cause your earlier session pinned is consistent with rivet-core/src/commits.rs: the trailer-grammar's is_artifact_id requires the prefix segment to be all uppercase letters with no digits, while the YAML side accepts any [A-Za-z][A-Za-z0-9]*-NNN-shaped id. So MAD1-101 round-trips fine through rivet validate (artifact id is legal), then gets rejected the moment a commit body cites it via Implements: MAD1-101 / Fixes: MAD1-101 / etc. That asymmetry is the friction you're hitting — the artifact id is created in one tool and rejected in another.

Two possible directions the routine sees, both reasonable:

1. Loosen the trailer parser to accept the same id grammar as the YAML side (allow digits in the prefix segment). One-line grammar change in commits.rs plus regression tests for MAD1-101, BUILD2-100, and the existing pure-alpha ids; should remain strict on truly malformed shapes (no leading digit, no empty prefix, etc.).
2. Tighten the YAML validator to reject prefix segments with digits at creation time (so rivet add / rivet validate flag MAD1-NNN before anyone tries to cite it in a commit). Safer long-term because the trailer grammar is also what tools downstream (rivet commits, dashboard) all share — but it's a breaking change for any project already using digit-bearing prefixes.

Both are tractable; the maintainer call is whether to align "down" (loosen commit) or "up" (tighten validate). Worth pinning before any branch opens.

To pick this up under the routine's "## Acceptance Criteria → branch → PR" workflow, the issue needs an explicit AC block on the body. Shape we look for (drop in / revise — feel free to pick one of the two directions above and trim the rest):

## Acceptance Criteria
- [ ] `rivet validate` and `rivet commits` agree on which prefix shapes are legal artifact ids — no id passes one and is rejected by the other.
- [ ] Regression fixture covers: `REQ-001` (alpha-only, must pass both), `MAD1-101` (alpha-then-digit prefix — outcome per direction chosen above), `1AB-001` (leading digit — must fail both), `-001` (empty prefix — must fail both).
- [ ] `cargo test -p rivet-core` green for the new fixtures; `cargo fmt --check` / `clippy --all-targets -- -D warnings` clean; `rivet validate` on this repo's own artifacts PASS unchanged.
- [ ] Commit trailers: `Fixes: REQ-NNN` (file a new REQ for "id grammar must be consistent across validate and commit-trailer parsing") + `Refs: #577`.
- [ ] **Explicitly out of scope (file separately):** changing id casing (e.g. allowing lowercase prefixes); changing the `-NNN` numeric-tail grammar.

Process side-rail (carried from every recent triage thread, mentioning because you may pick the answer up in a comment thread): https://pulseengine.eu/blog/ — the binding workflow guidance — is currently HTTP 503 and has been across 30+ consecutive routine runs (TLS-expiry / site-infra symptom). The routine can't open a PR without consulting it first, so even with AC in place this will land as a draft pending blog access — consistent with the posture on PRs #525, #537, #539, #541. The cert renewal is the unblocker.

Not implemented this run.

---

Generated by Claude Code — issue-triage agent run 2026-06-24.

---

Generated by Claude Code

[avrabe]

Triage (scheduled run): tool-friction label is already correct — no label change.

Body has good root-cause analysis but no explicit acceptance criteria, so I'm not starting implementation yet. To make this agent-closable, please add a section like:

## Acceptance Criteria
- [ ] Pick the direction: tighten `rivet validate` to reject `MAD1-101`-style IDs, OR loosen `commits.rs::is_artifact_id` to accept them
- [ ] `rivet validate` and the commit-trailer parser agree on the same ID grammar (assertion: a unit test feeds the same set of IDs to both and gets identical accept/reject verdicts)
- [ ] Existing valid IDs in the repo (e.g. `UDS-102`, `TRACE-104`, `BUILD-100`) remain accepted; the chosen verdict for digit-bearing prefixes (`MAD1-101`) is documented
- [ ] CHANGELOG entry under `[Unreleased]`

The first bullet is the design decision that has to land before implementation; once you pick a direction, the rest is mechanical.

— rivet triage (Claude Code, scheduled)

---

Generated by Claude Code