deconstruct-sat-solver

Task gallery / Security / cyber / Version 1

Reference solution (oracle): Unverified
No-op resistance check: Unverified
QA verdict: Not recorded
Aggregate eligibility: Eligible version

The task

Deconstruct a SAT Solver /app/oraclebinary/cadical is a protected, deterministic reference program: you can run it freely, but its implementation is not readable. Reimplement it. Its observable surface is a no-argument, one-physical-line-per-command CNF protocol with solve (NVARS <= 512), minmodel (<= 120), count (<= 24), and backbone (<= 24), plus --help/-h, --version/-V, and invalid argument handling. That binary …

Excerpt only; not a complete executable task specification. Task ID task_eb165d4186e84dec · version 1. Selection status: not fully recorded.

Results by model and effort

This exact task version only. Every rate is passes / valid runs with a Wilson 95% interval; n≤2 remains counts only. Macro aggregation does not apply within a single task. Baselines are summarized in the quality gates, not ranked. A passed no-op resistance check means that doing nothing did not solve the task.

Current evidence

Model / effort	Verifier result [95% Wilson]	Diagnostic labels / valid	HARNESS_ERROR · excluded	Outcome classes
Claude Opus 5 · default (unverified)	0/5 · 0.0% [0.0%, 43.4%]	0 / 5	0	GOOD_FAILURE: 0 BAD_FAILURE: 0 GOOD_SUCCESS: 0 BAD_SUCCESS: 0

Legacy evidence

Model / effort	Verifier result [95% Wilson]	Diagnostic labels / valid	HARNESS_ERROR · excluded	Outcome classes

Difficulty, with uncertainty

Selected-best difficulty: 100.0% [56.6%, 100.0%]. This is one minus the best measured current-row pass rate among rows with at least five valid runs.

The maximum is selected from the observed rows, so the estimate is optimistic. Its Wilson interval is not adjusted for model selection. “Hard” requires qualifying GOOD_FAILURE evidence; the threshold flag additionally requires the interval and recorded quality gates to meet the documented rule. This is not independent validation. Threshold flag: no.

Every recorded run

5 recorded trials, including controls. Diagnostic labels do not replace verifier reward. Labels and quality checks are source declarations, not independently attested here. Internal evidence remains private; a missing public URL is shown explicitly. Records with invalid or absent rewards are not scored.

Run / model	Date / source	Reward / eligibility	Diagnostic class / source	Resources	Evidence
`trial_9ae7e9c1ba274333` Claude Opus 5 · default (unverified)	Oct 9, 2026 console	0 · valid	No diagnostic label derived	Tokens: 117,456 Turns: 84 Cost: $9.0410	Not publicly available
`trial_14ae34bdaa994df3` Claude Opus 5 · default (unverified)	Oct 9, 2026 console	0 · valid	No diagnostic label derived	Tokens: 140,674 Turns: 102 Cost: $12.2750	Not publicly available
`trial_6396bfc3f4ea4c15` Claude Opus 5 · default (unverified)	Oct 9, 2026 console	0 · valid	No diagnostic label derived	Tokens: 115,004 Turns: 69 Cost: $7.8494	Not publicly available
`trial_a8143158c8e9406d` Claude Opus 5 · default (unverified)	Oct 9, 2026 console	0 · valid	No diagnostic label derived	Tokens: 108,898 Turns: 82 Cost: $8.0682	Not publicly available
`trial_e1868caf1b2a45f9` Claude Opus 5 · default (unverified)	Oct 9, 2026 console	0 · valid	No diagnostic label derived	Tokens: 143,378 Turns: 102 Cost: $11.7523	Not publicly available