source: cell_forensics.csv; rollup.json

Task registration facts

The accession is H-M027, the identifier assigned to this maintenance task, and the host-side record of the task names it duplicating a deliverable document. The task index records the task as registered for measured runs. An arm, one software build supplied to an agent, ran as H-LAP-L1, H-LAP-L1H, H-LAP-L2C, H-LAP-L2S, H-LAP-L2T, H-LAP-L3B, H-LAP-L5P, H-NON, H-STR. A cell, one execution for an arm and seed, is the row counted below. A seed, the repeat number that distinguishes repetitions, is part of the cell identifier. A run set, one named batch of cells, is part of the evidence record. A probe, a hidden check used to score work, can appear in a run-set identifier. A documentation dose, the amount of documentation supplied with a build, can appear in a run-set identifier. An agent-coder, a local model alias used for a coding run, can appear in a run-set identifier (arm identifiers). source: 5. Experiment/4. Task Library/Project H/H-M027/task_manifest.json; 5. Experiment/4. Task Library/Project H/H-M027/README.md; cell_forensics.csv

Attempt outcomes by arm

A hidden tier, host-side checks withheld from the agent, decides the hidden result. A visible tier, checks shown during the attempt, can disagree with it. An invalid attempt, an attempt carrying one or more instrument flags, remains visible in its outcome count.

ArmAttemptsHidden-tier passHidden-tier failNot runInvalid
H-LAP-L1116324
H-LAP-L1H106312
H-LAP-L2C11000
H-LAP-L2S11001
H-LAP-L2T11000
H-LAP-L3B11001
H-LAP-L5P127505
H-NON86201
H-STR146624
source: cell_forensics.csv

Recorded failure causes

Failure causeCountShare of task rowsExample cells
A11, a cause code (a grouped failure label) for no success, visible verdict varied between passes11.7% of 59 task rows041-20260829-lap-1-span-agentcoder-first-contact / H-LAP-L1__H-M027__full__s1
A2, a cause code (a grouped failure label) for success after a hidden-failure retry46.8% of 59 task rows170-20260911-auto-igx-thor-agent-coder-1308 / H-STR__H-M027__full__s5; 189-20260912-auto-dgx-agent-coder-0212 / H-LAP-L1__H-M027__full__s7; 193-20260912-auto-igx-thor-agent-coder-0343 / H-LAP-L1H__H-M027__full__s8
A3, a cause code (a grouped failure label) for success after a visible-failure retry23.4% of 59 task rows060-20260902-nd-haiku-hint-off / H-STR__H-M027__full__s1; 061-20260902-nd-haiku-hint-on / H-LAP-L5P__H-M027__full__s1
A4a, a cause code (a grouped failure label) for same requirement failed on every pass35.1% of 59 task rows063-20260902-nd-local-feedback-off / H-STR__H-M027__full__s1; 100-20260910-auto-dgx-agent-coder-0020 / H-LAP-L1H__H-M027__full__s4; 173-20260911-auto-igx-thor-agent-coder-1507 / H-STR__H-M027__full__s6
A5, a cause code (a grouped failure label) for no edit made1016.9% of 59 task rows053-20260831-survey-qwen3coder-throughput / H-LAP-L5P__H-M027__full__s1; 095-20260909-auto-dgx-qwen3-4b-fp8-2125 / H-LAP-L1H__H-M027__full__s3; 114-20260910-auto-igx-thor-qwen3-4b-fp8-1535 / H-LAP-L1__H-M027__full__s4
A7, a cause code (a grouped failure label) for visible failure on every pass23.4% of 59 task rows060-20260902-nd-haiku-hint-off / H-LAP-L5P__H-M027__full__s1; 123-20260910-auto-dgx-agent-coder-2016 / H-NON__H-M027__full__s3
source: rollup.json; cell_forensics.csv

Visible and hidden tier disagreements

Run setCellFieldExpected recordObserved record
203-20260912-auto-igx-thor-agent-coder-1053H-LAP-L5P__H-M027__full__s7population_keyregistermissing_from_forensics (record comparison)
source: disagreements.csv

Agent pass timing figures

The pass evidence contains no duration field, so it provides no median or longest agent pass figure. source: pass_forensics.csv

Evidence sources

The numbers above come from the named forensics files and the task manifest. The forensics package records provenance commit 6a0e5432f8750522e3bbacaf89b48de13a8e8058. source: cell_forensics.csv; pass_forensics.csv; disagreements.csv; rollup.json; 5. Experiment/4. Task Library/Project H/H-M027/task_manifest.json; provenance.json