smoke2: scout clean-context synthesis
completed- finished
- 2026-07-02
- took
- 9m 0s
- fired by
- m5Max
- wf
- #196
Runs
The squad's recent workflow runs, straight from the harness — failures included. Every run below was fired, executed by the local agents, and graded by deterministic gates that no agent can talk its way past. A red row is a gate doing its job.
199
workflows fired
145
completed
51
failed
since 2026-06-09 · as of 2026-07-03 19:51 UTC · refreshed at each deploy
23 of the last 40 finished runs are withheld from this log: security-sensitive work isn't published until it's merged and reviewed. Withheld runs still count in the totals above.
consistency scan: three-language house-style survey
failedscout×3
wall-timeout — ran out of clock
- finished
- 2026-07-02
- took
- 17m 0s
- fired by
- m5Max
- wf
- #194
interp recon retry: tight MLX lane + synthesis
completed- finished
- 2026-07-02
- took
- 13m 50s
- fired by
- m5Max
- wf
- #193
interp recon: MLX tooling + probes/steering + quantization
failedwall-timeout — ran out of clock
- finished
- 2026-07-02
- took
- 17m 0s
- fired by
- m5Max
- wf
- #192
out-of-band flags: terminal-status regression tests
completed- finished
- 2026-07-02
- took
- 4m 58s
- fired by
- m5Max
- wf
- #191
verdict-no-change: step4 terminal-status tests (re-fire)
failedefficacy gate failed — the deterministic gate outranks any verdict
- finished
- 2026-07-02
- took
- 9m 18s
- fired by
- m5Max
- wf
- #190
verdict-no-change: step4 runner tests + step5 full review
failedbrain failed mid-run
- finished
- 2026-07-02
- took
- 13m 5s
- fired by
- m5Max
- wf
- #189
verdict-no-change: fix evidence suffix-match boundary hole
completed- finished
- 2026-07-02
- took
- 6m 11s
- fired by
- m5Max
- wf
- #188
verdict-no-change: step1 verdict.py + tests
failedefficacy gate failed — the deterministic gate outranks any verdict
- finished
- 2026-07-02
- took
- 17m 42s
- fired by
- m5Max
- wf
- #187
compose: determine if the tok/s and w/tok meters
failedno edit made on a fix brief
- finished
- 2026-07-02
- took
- 33m 6s
- fired by
- Gilbert
- wf
- #186
Cockpit Step 4 — Approval Bell
failedefficacy gate failed — the deterministic gate outranks any verdict
- finished
- 2026-07-02
- took
- 50m 54s
- fired by
- kira
- wf
- #185
cockpit-slice-step3-toolcall-cards
completed- finished
- 2026-07-02
- took
- 37m 58s
- fired by
- kira
- wf
- #184
cockpit-step2-risk-wave-panel
completed- finished
- 2026-07-02
- took
- 18m 52s
- fired by
- kira
- wf
- #183
cockpit-slice-step1-live-gauges
completed- finished
- 2026-07-02
- took
- 13m 27s
- fired by
- kira
- wf
- #182
GLM game-dev probe: lightweight level-based mini-game (design-first runnable)
failedwall-timeout — ran out of clock
- finished
- 2026-07-02
- took
- 31m 21s
- fired by
- m5Max
- wf
- #180
motu-golden-path-combat-fix
failedno edit made on a fix brief
- finished
- 2026-07-02
- took
- 43m 19s
- fired by
- m5Max
- wf
- #179
motu A2c: strengthen the capstone assertion (finish #175 — rewire done, one gap)
completed- finished
- 2026-07-01
- took
- 21m 37s
- fired by
- m5Max
- wf
- #177
the lab substrate workflow log, sanitized (allowlisted fields; security-sensitive runs withheld until merged + reviewed). Failure labels are fixed machine classes — raw run output never leaves the lab.