github-devloop implementation for #367

[ElonSG]

github-devloop implementation PR for issue #367

[ElonSG]

github-devloop PR child open

[ElonSG]

github-devloop PR is ready for review

[ElonSG]

github-devloop PR is ready for review

[ElonSG]

github-devloop PR review decision: approve
Verdicts: teleology=approve parsimony=approve fidelity=approve natural-ownership=approve proportional-containment=approve

teleology:
The diff cleanly implements the established failure-versus-error taxonomy: engine assertions create a private structural marker, report construction emits assertion_failure only when that marker survives the mlua chain and otherwise emits neutral test_error, and the regression tests verify text-independent classification plus legacy fields and exit behavior at crates/fkst-framework/tests/test_runner_cli.rs:1897 and :1970. Documentation at SPEC.md:118 keeps downstream retry policy outside the engine. No evidenced BEAUTY-GATE smell blocks advancement.

parsimony:
The diff is parsimonious and satisfies the settled requirements: assertions acquire a private structural marker at crates/fkst-framework/src/test_assertions.rs:94, classification occurs once at crates/fkst-framework/src/test_runner.rs:261, load failures directly use test_error at crates/fkst-framework/src/test_runner.rs:139, and the regression suite proves prose independence and legacy serialization at crates/fkst-framework/tests/test_runner_cli.rs:1931 and crates/fkst-framework/tests/test_runner_cli.rs:1970; no magic number, proxy classification, or symptom branch remains.

fidelity:
The diff faithfully implements the stated requirement: engine assertions carry a structural marker, mlua wrapping is traversed by source chain, load and unmarked runtime failures remain neutral test_error, and end-to-end tests verify that error prose cannot control classification while legacy fields and exit behavior remain unchanged. ⟦AI:FKST⟧

natural-ownership:
The change places the invariant at its natural owner: assertions acquire a private structural marker at creation (test_assertions.rs), and the runner derives failure_kind from the typed mlua cause chain before serialization (test_runner.rs). The diff directly verifies prose independence and both failure kinds (test_runner_cli.rs), preserves the legacy fields through an exact golden projection (test_runner_cli.rs), and documents that retryability remains downstream (architecture.md). No evidenced BEAUTY-GATE smell blocks advancement.

proportional-containment:
The diff cleanly implements the established failure-versus-error taxonomy at the producer boundary: engine assertions receive structural provenance, all other failures remain neutrally named test_error, retry policy stays downstream, and regression tests verify text independence, legacy serialization, pass serialization, and unchanged exit behavior. The documentation explicitly preserves this containment at docs/architecture.md:115; no BEAUTY-GATE smell or stated-requirement gap is evidenced. ⟦AI:FKST⟧

⟦AI:FKST⟧

[ElonSG]

github-devloop is merging PR #369

[ElonSG]

github-devloop merged PR #369