claw/docs/superpowers/reports/2026-04-19-timeout-budget-rerun-hygiene-report.md

Timeout Budget and Rerun Hygiene Report

Date: 2026-04-19 Plan: docs/superpowers/plans/2026-04-19-timeout-budget-rerun-hygiene-plan.md Source diagnostic: tests/fixtures/generated_scene/timeout_regression_diagnostic_2026-04-19.json

Scope

This plan only converts the three timeout diagnostic records into a rerun-hygiene layer.

It does not:

1. change analyzer or generator code
2. update scene_execution_board_2026-04-18.json
3. promote scenes
4. rerun the full 102 sweep
5. treat rerun success as validated scene pass

Frozen Diagnostic Input

Diagnostic label	Count
timeout-rerun-pass	2
timeout-rerun-fail-closed	1
Total	3

Hygiene Mapping

Diagnostic label	Hygiene status
timeout-rerun-pass	rerun-resolved-pass
timeout-rerun-fail-closed	rerun-resolved-fail-closed
timeout-rerun-timeout	rerun-still-timeout
timeout-rerun-error	rerun-error
timeout-large-source	rerun-still-timeout
timeout-command-hang	rerun-still-timeout
timeout-nondeterministic	rerun-error
timeout-source-scan-heavy	rerun-still-timeout
timeout-unknown	rerun-error

Hygiene Results

Scene id	Scene	Diagnostic label	Hygiene status	Elapsed seconds	Result
sweep-015-scene	????	timeout-rerun-pass	rerun-resolved-pass	74.76	readiness A
sweep-025-scene	??????????	timeout-rerun-pass	rerun-resolved-pass	49.03	readiness A
sweep-040-scene	?????	timeout-rerun-fail-closed	rerun-resolved-fail-closed	45.91	fail-closed, readiness C

Summary

Hygiene status	Count
rerun-resolved-pass	2
rerun-resolved-fail-closed	1
rerun-still-timeout	0
rerun-error	0

Interpretation

1. The timeout bucket should not be treated as a single unreadable category.
2. Two timeout records are budget-sensitive pass candidates under the bounded rerun budget.
3. sweep-040-scene should not stay grouped with hard unreadable inputs. Under rerun hygiene, it belongs to rerun-resolved-fail-closed, which means it should feed the structured fail-closed layer rather than the unreadable bucket.
4. This remains a hygiene layer only. None of these three records are promoted or merged into the execution board by this report.

Recommendation

Use the rerun hygiene layer before any timeout implementation work or execution-board update.

The highest-value follow-up is to apply this hygiene rule to future sweeps so budget-sensitive scenes are separated into:

1. rerun-resolved-pass
2. rerun-resolved-fail-closed
3. rerun-still-timeout
4. rerun-error

instead of collapsing all timeout outcomes into source-unreadable.