Upstream results
The indexed record of reports, reviewed fixes and their current status.
§ 3.7
External subjects and upstream results
The 17 records in this upstream index concern systems Omega did not build. Eight are defects Omega reported and upstream contributors fixed; four are fixes Omega wrote and maintainers reviewed and merged; one are pull requests Omega authored and still open; four are reports without a merged fix. None of this is an endorsement, and no logo is needed to say it.
| External subject | What was tested | What was found and fixed | Upstream status |
|---|---|---|---|
| Inspect AI (UK AI Security Institute evaluation framework) | What the completed-sample count in a finished run's metadata actually counts when a scorer raises. | completed_samples was computed from scored samples, so the same failing run reported 6 of 6 or 4 of 6 depending only on scorer declaration order. Issue #4602, 23 July 2026. Pull request #4604 by Anai-Guo, merged 29 July 2026; released in 0.3.252 (4 August 2026). |
RELEASED Reported by Omega; fixed by another contributor and merged by the maintainers. fix · reportfinding |
| LemmaScript (a verified-language toolchain emitting Dafny and Lean) | Whether cross-file externs are modelled as deterministic functions, so two calls to a random source prove equal. | Every called cross-file symbol was emitted as an axiomatised pure function; dafny verify reported 1 verified, 0 errors on code whose rollDie() returns different values per call. Issue #206, 24 August 2026. Pull request #207 (impure extern) by the maintainer, merged within six hours; tagged v0.6.1 the same day. |
RELEASED Reported by Omega; fixed by the maintainers. fix · report |
| inspect-robots (robocurve embodied-evaluation harness) | Whether the persisted log distinguishes an embodiment-adopted verdict from one sampled from a vision-language model. | Neither the final log nor the live snapshot recorded which path produced each operator judgement, so verdicts from different paths were indistinguishable on disk. Issue #413, 27 August 2026. Pull request #422 by the maintainer, merged 31 August 2026; released in v0.58.0 (2 September 2026). |
RELEASED Reported by Omega; fixed by the maintainers. fix · report |
| NVIDIA-NeMo labs-OO-Agents | Where a tracing span for a generator-bodied agent method begins and ends. | Generator and async-generator methods were routed to the coroutine wrapper, so generation was skipped and the span ended at generator creation rather than at consumption. Issue #38, 27 July 2026. Pull request #120 (fixes #38), merged 31 August 2026; released in v0.0.10 (4 September 2026). |
RELEASED Reported by Omega; fixed by the maintainers. fix · report |
| NVIDIA SkillSpector | Whether the Anthropic provider's structured-output schema is accepted by the API it targets. | Every LLM-mode scan through the Anthropic provider failed: the schema carried minimum and maximum bounds the structured-output endpoint rejected. Issue #76, 15 June 2026. Closed by the maintainers as resolved by pull request #85 (native Anthropic provider), merged 16 June 2026; first included in tagged release v2.5.0 (27 July 2026). The change does not edit the bounds directly; the maintainers' closing comment is the link between report and fix. |
RELEASED Reported by Omega; fixed by the maintainers. fix · report |
| cedar-policy cedar-spec (Cedar's Lean model and differential tests) | Whether the Rust and Lean typecheckers compute the same typed expression, not only the same pass/fail verdict. | Eight phantom-divergence classes surfaced while building the target; the pull request adds a differential target over the typed expression. Pull request #995 by Omega, 28 July 2026, against issue #840 (maintainer-filed). Open pull request. A maintainer requested changes on 28 July; Warren's latest revision was 24 August; no later maintainer review as of 20 September 2026. |
OPEN PULL REQUEST A pull request Omega authored; awaiting upstream action. pull request |
| pulp-platform common_cells (an open hardware IP library) | Whether the ECC encoder and decoder pair satisfy the four properties documented in the decoder header, at every data width. | The proof went through, and a cover requirement on the decoder's syndrome output showed that the output was declared and never driven. A one-line assignment fixed it; the proof and a width sweep to 64 bits passed with the fix. Pull request #355 by Omega, 11 August 2026, as a subtask of issue #288 (maintainer-filed); the undriven output was found on 3 September 2026 while adding the coverage the maintainer asked for. Reviewed and merged by the maintainer on 3 September 2026. On the default branch; in no tagged release as of 20 September 2026. |
MERGED Fix authored by Omega, reviewed and merged by the maintainer. fixfinding |
| inspect-robots (robocurve embodied-evaluation harness) | Whether the persisted run specification records which grader was configured and the configuration governing it. | The specification carried the policy and embodiment configuration but nothing about the grader, so two runs graded by different models against different rubrics wrote byte-identical specifications. The pull request adds the grader name and its resolved configuration. Pull request #425 by Omega, 31 August 2026, the follow-up the maintainer invited on issue #413. Reviewed and merged upstream on 2026-09-21. Merge commit 79c782c3e380f462a4369ff7cf2c876416456a34. |
MERGED Fix authored by Omega, reviewed and merged by the maintainer. fix |
| Microsoft Agent Framework (Python) with its agent-hooks enforcement bundle | Whether the documented enforcing configuration mediates every side effect observed out of process, and what signal exists when a delegated child agent or a later-placed middleware sits outside it. | The documented configuration behaved correctly: every side effect observed out of process had a causally preceding permit for exactly that operation. Two documented configuration requirements, a bundle on each delegated child agent and the bundle first in the middleware list, produce no signal when violated. This is not a vulnerability report: the framework does not claim complete mediation, and the request is for a signal, not for a change in what is enforced. Issue #8050, 4 September 2026. Open issue. A maintainer replied on 4 September 2026 that guarding a delegated agent is the agent author's responsibility and that they do not expect to change this; no later maintainer response as of 20 September 2026. |
OPEN ISSUE Reported by Omega; no merged fix. issue |
| PyBaMM (an open-source battery modelling library) | Whether the BasicDFN model conserves lithium when the transference number depends on concentration. | An independently computed lithium inventory drifted by 2.44% on every revision and at every refinement, matching the rate predicted from the missing transference-gradient term; a project contributor confirmed the missing term on the parent thread and asked for a separate report. Issue #5745, 6 September 2026, split from #5700 at a contributor's request. Warren's pull request #5747, covering the lithium and sodium BasicDFN variants, was approved and merged by the maintainer on 18 September 2026. On the default branch; in no tagged release as of 20 September 2026. |
MERGED Fix authored by Omega, reviewed and merged by the maintainer. fix · reportfinding |
| PyBaMM (an open-source battery modelling library) | Whether the half-cell diagnostic Total lithium in electrolyte uses the electrolyte's own thickness. | It multiplies by the full-cell thickness (767 µm in the Xu2019 parameter set) where the electrolyte occupies 67 µm, overweighting the term by a factor of 11.45; a project contributor confirmed it and asked for a separate report. Issue #5746, 6 September 2026, split from #5700 at a contributor's request. Maintainer-authored pull request #5765 corrects the half-cell lithium diagnostic and adds regression tests. Merged 2026-09-22; merge commit 9daad126258db8f6e621e4746cd3bdac062e7f1f. |
MERGED Reported by Omega; fixed by the maintainers. fix · report |
| pandapower (an open-source power-system analysis library) | Whether transformer vector groups the three-phase power flow does not support are rejected rather than solved. | Four of fourteen unsupported groups returned converged=True with no warning and identical non-physical results; ten raised the intended error; one group documented as supported is rejected by the code. Issue #3122, 6 September 2026. Open issue. No maintainer response as of 20 September 2026. |
OPEN ISSUE Reported by Omega; no merged fix. issuefinding |
| cvc5 (an SMT solver) | Whether an emp-free separation-logic formula with magic wand is classified consistently with its heap semantics. | cvc5 returned sat for a formula whose spatial constraints require incompatible heaps. An upstream contributor reproduced the result and identified a polarity-handling cause. Issue #12913 by Omega, 5 September 2026. Open issue. An upstream contributor confirmed the root cause on 6 September and opened draft pull request #12925; no merged fix as of 20 September 2026. |
OPEN ISSUE Reported by Omega; no merged fix. issue |
| cvc5 (an SMT solver) | Whether sep.emp is cached with the polarity needed by the separation-logic solver. | The cached polarity could be wrong, producing an incorrect result on the upstream reproducer. The proposed fix changes the cache key and adds regression tests. Pull request #12937 by Omega, 11 September 2026, against issue #12870. Reviewed and merged upstream on 2026-10-01. Merge commit 7467ffe6f725e7e83c3f67b4edd407fcb71b1367. |
MERGED Fix authored by Omega, reviewed and merged by the maintainer. fix · report |
| mpmath (an arbitrary-precision numerical library) | Whether interval exp encloses independently computed endpoint values. | mpmath's interval exponential excluded a high-precision endpoint value on a public reproducer; independent Decimal and Arb computations agreed on the exclusion. A maintainer confirmed the problem. Issue #1181 by Omega, 14 September 2026. Open issue. A maintainer's attempted pull request #1183 was closed unmerged after additional counterexamples showed the approach was incomplete; no merged fix as of 20 September 2026. |
OPEN ISSUE Reported by Omega; no merged fix. issue |
| con-leche (a Lean proof-checking tool) | Whether the command fails when its checker does not actually run. | The command could exit successfully without invoking the checker, so automation could accept a non-check as a pass. Issue #8 by Omega, 10 September 2026. The maintainer fixed the exit code directly on the default branch in commit 19a2f243 on 10 September 2026. No tagged release exists as of 20 September 2026. |
MERGED Reported by Omega; fixed by the maintainers. fix · report |
| Orbit (an agent runtime) | Whether the documented Anthropic provider works from the project's declared installation dependencies. | The provider imported the Anthropic SDK, but the package did not declare that dependency, so a clean documented installation failed at import time. Issue #1 by Omega, 23 July 2026. Pull request #8 by the maintainer, merged 17 September 2026; released in v1.0.1 the same day. |
RELEASED Reported by Omega; fixed by the maintainers. fix · report |
Released means the fix is in a tagged release. Merged means it is on the default branch but not verified in a tagged release. Open pull request means an Omega-authored change awaits upstream action. Open issue means a report remains open without a merged fix. Maintainer acceptance is review of one change, not endorsement of Omega.