Integration testing — v0.13.1

Session run: 2026-09-27 · against develop at b0b12d5c (post-freeze). Result: PASS — clear to release.

Features exercised

None. v0.13.1 ships no new language-visible feature: no grammar change (beyond two architecture-evidence comment markers), no reference/error-codes.md change, no reference/spec/ change. Everything since v0.13.0 is compiler-internals work (the closing slice of metel-core#1047, the binding-identity/resolved-IR unification), new frontend tooling APIs consumed by metel-lsp (a separate crate, not the language surface), and internal engineering documentation (the Architecture Atlas). See release-notes/changelog.md's ## v0.13.1 section for the fix and API list.

Interaction matrix

Not applicable — an N×N matrix requires at least two features to pair, and none shipped this milestone. This is the "genuinely no cross-feature risk" case this report's own template anticipates: still produced, still dated, so /cut-release has its gate artifact.

Fixtures added

None — nothing to interact-test.

Findings

(none)

Sign-off

No high-risk pairs exist (no features shipped to pair). Full release suite green on b0b12d5c:

• cargo test --release (metel-interpreter, all crates): 0 failures.
• cargo clippy --release --lib -- -W clippy::pedantic: clean.
• cargo build --release -p metel: clean.
• move-check-count: skipped_generic_bodies_user_total=15, at/below the documented baseline of 18 (an improvement, consistent with metel-core#1226's fix). user_move_violations=70, entirely from the corpus's deliberate negative move-check fixtures (*neg_*.mtl, evaluator/move_check/*.mtl) whose purpose is to trigger exactly this.
• check_doc_examples.py against README.md, docs/getting-started/tutorials, docs/reference/spec: 195 passed, 0 failed, 48 skipped (all with a recorded skip reason: unverifiable fragment or an explicit skip annotation).

Clear for /cut-release v0.13.1.