Cases

Operating episodes

One concrete episode at a time: what triggered it, how the system responded, and what outcome followed.

Start Here

The clearest first episodes for seeing how the current system behaves under real pressure.

Panic-prone shortcut blocked before write

A repository-written `expect(...)` shortcut in the websocket stream reached the pre-write lane and was blocked through `RUST-ANTI-PANIC-001`, with a concrete rule, file path, and remediation before the write could land.

Trigger: Repository-written `expect(...)` introduced a panic edge into shared Rust code.

Response: The pre-write lane blocked the write, named the file, and attached remediation.

Outcome: The shortcut never entered the tree as accepted code.

Open case

Sleep-based timing shortcut blocked before write

A real `tokio::time::sleep(...)` call in a file-search test reached the pre-write lane through `RUST-ANTI-TESTSLEEP-001`, with remediation pointed toward paused time or an injected clock.

Trigger: A direct `tokio::time::sleep(...)` call introduced nondeterministic time control into repository test code.

Response: The pre-write lane blocked the write and redirected the fix toward deterministic time control.

Outcome: The sleep-based workaround was rejected before it could settle into normal test style.

Open case

Smoke replay exposed a named snapshot mismatch

A smoke replay did not collapse terminal drift into a generic suite failure: it named the exact snapshot path and printed a unified diff directly in the terminal.

Trigger: The live terminal snapshot diverged from the saved expected state for one smoke replay step.

Response: The replay harness named the snapshot path and surfaced a concrete diff in terminal output.

Outcome: UI drift became one reproducible debugging surface instead of a generic replay failure.

Open case

More Cases

The rest of the current published operating episodes.

Serial test attribute blocked before write

A `#[serial]` shortcut in a headless login test was blocked on the pre-write lane through `RUST-ANTI-TESTSERIAL-001`, before shared-state coordination could normalize as test style.

Trigger: `#[serial]` introduced a shared-state escape hatch into a repository test.

Response: The pre-write lane blocked the write and tied the reject to the exact file and snippet.

Outcome: The attribute was stopped before it could become normal test style.

Why it matters: The repo rejected the shortcut at the moment of write instead of asking later reviewers to infer the discipline from downstream flakes.

Open case

Serial test attribute blocked before write

False config-load rejection captured as typed errata

A stale hook binary falsely rejected `.tsqoba/tsqoba.toml`, and the incident was recorded as typed errata and fixed instead of being left as unexplained tool noise.

Trigger: An outdated hook binary mis-modeled repo config and raised a false parse error.

Response: The block was captured as typed errata, explained, and traced to stale hook tooling rather than bad repo config.

Outcome: The incident moved from confusing tool noise to recorded follow-up with a concrete fix.

Why it matters: Dogfooding only helps if bad rejections are recorded, explained, and fixed instead of lingering as unexplained tool noise.

Open case

Text-first

TEXT

HOOK-RUST-CHECK-CONFIG-LOAD-001
failed to parse TOML
line 28, column 2
unknown field 'audit'
28 | [audit]