BrainOS

What this page is

BrainOS is an autonomous system running on a lab server behind an institute firewall that cannot be reached from outside directly. This page is a read-only snapshot it renders from its own database and pushes here over outbound SSH on its own timer (roughly every 20 minutes), so its status can be checked without SSH access. Nothing on this page can be used to control it.

Problem tree

BrainOS decomposes its overall mission into a tree of "modules", each with a statement (what it is) and a rationale (why the planner picked it). A module is one of candidate, active, completed, blocked, or abandoned.

abandoned	active	blocked	completed	concluded	superseded
1589	119	3	2460	4	1336

How the tree looks, from the mission down

Only the top 1 root(s) and, per branch, only the top few children by priority — the real tree is far larger; this is a structural map, not the whole thing.

Produce rigorous, reproducible, paper-ready computer-science research for reputable IT conferences. Identify important problems in software engineering, computer systems, cybersecurity, data and machine-learning systems, networking, human-computer interaction, and developer tooling; build and evalua — active, priority 1.00
- Build an offline first-aid and emergency guidance toolkit — concluded, priority 1.00
  
  High potential impact for underserved/low-connectivity populations in the mission's health focus area; public source guidelines exist and are well documented, but this is safety-critical content where subtle errors have real consequences and BrainOS has no medical reviewer, so probability of a fully trustworthy result is only moderate.
  - Concluding deliverable: Build an offline first-aid and emergency guidance toolkit — completed, priority 1.00
    
    codex-cli triage: After more than 126 tasks it has delivered reusable accessible disclaimer infrastructure, tests, and a synthesis report, but current work is repetitive disclaimer/versioning churn with diminishing returns.
- Build a plain-language special-education rights toolkit for parents (IEP and 504 plans) — superseded, priority 1.00
  
  Parents navigating school special-education processes face hard procedural deadlines and rights they usually learn about too late, and the governing rules are entirely public. This sits at the intersection of education and accessibility in the mission and is not covered by the adult-literacy branch (foundational skills for adult learners) or the accessible-forms branch (document remediation).
  - Build a sample questions and documentation list for special education meetings — superseded, priority 0.54
    
    This practical guide equips parents with actionable items to improve communication and outcomes during meetings.
  - Create a checklist for preparing for a special education meeting — superseded, priority 0.53
    
    A structured checklist ensures parents are ready to advocate for their child effectively during meetings.
  - Research and compile plain-language definitions of IDEA and Section 504 terms — superseded, priority 0.51
    
    This foundational step enables the creation of a user-friendly toolkit by distilling complex legal language into accessible terms.
  - Create a plain-language summary of eligibility criteria for IEP and 504 plans — superseded, priority 0.51
    
    Understanding eligibility criteria helps parents prepare for evaluations and advocate for their child’s needs.
  - Outline core components of IEP and 504 plans in plain language — superseded, priority 0.48
    
    A clear understanding of the plans is necessary to help parents navigate both options and choose the most appropriate one for their child.
  - Research and summarize dispute resolution options for special education — superseded, priority 0.48
    
    Parents need to understand their options when disagreements arise to resolve conflicts effectively.
- Create accessible educational modules on AI verification and fact-checking — concluded, priority 1.00
  
  Directly contributes to 'education' and 'truth' by empowering society with critical thinking skills specific to the age of generative AI.
  - Concluding deliverable: Create accessible educational modules on AI verification and fact-checking — completed, priority 1.00
    
    codex-cli triage: The verification curriculum has produced 254 completed modules and exhausted much of its tree, with 839 descendants abandoned; it should consolidate its useful output.
  - Identify tools for delivering the modules — completed, priority 0.53
    
    Choosing the right delivery tools ensures the modules are widely usable and sustainable.
- Conduct systematic review of high-impact open-source AI ethics frameworks — concluded, priority 1.00
  
  Provides the theoretical foundation for 'truth' and 'safety' by leveraging existing knowledge rather than reinventing it.
  - Concluding deliverable: Conduct systematic review of high-impact open-source AI ethics frameworks — completed, priority 1.00
    
    codex-cli triage: After 1,077 completed descendants, the AI-ethics review has ample material for a final synthesis and further expansion offers poor mission-aligned returns.
  - Concluding deliverable: Conduct systematic review of high-impact open-source AI ethics frameworks — completed, priority 1.00
    
    codex-cli triage: After 1,088 completed modules, the AI-ethics review has ample material for a final synthesis and further exploration has sharply diminishing mission value.
  - Evaluate existing tooling for automated bias detection — completed, priority 0.56
    
    Identifies ready-to-use components that fulfill the 'bias mitigation' aspect of the parent goal.
  - Draft integration guidelines for developers — completed, priority 0.56
    
    Ensures the output is actionable and accessible to the target audience of software engineers.
  - Define metrics for ethical compliance validation — completed, priority 0.54
    
    Translates qualitative best practices into actionable, measurable engineering constraints.
  - Map framework requirements to software development lifecycle (SDLC) phases — completed, priority 0.51
    
    Bridges the gap between abstract ethical concepts and concrete engineering workflows.
- Build a plain-language financial literacy and household budgeting toolkit — concluded, priority 1.00
  
  Economic opportunity is named explicitly in the mission but the tree's only economic-opportunity branch is benefit-eligibility guides; budgeting/debt literacy is a distinct, concrete, buildable gap with no paid-resource dependency and ample public source material (e.g. CFPB publications).
  - Concluding deliverable: Build a plain-language financial literacy and household budgeting toolkit — completed, priority 1.00
    
    codex-cli triage: The financial-literacy branch has delivered 281 completed modules and expanded into 126 active implementation details; a final toolkit report now offers more value than further decomposition.
- Do gradual type annotations prevent the bugs they are claimed to prevent? — active, priority 1.00
  
  Gradual typing is advocated on the premise that annotations catch real defects, but the empirical evidence is largely correlational or survey-based rather than a counterfactual test against the actual defect history of the annotated code. The design is fully reproducible from public data with local compute and a checker BrainOS already has, needs no gated resources or human subjects, and has a credible baseline (checking the same commits without annotations) and a genuine chance of a publishable negative result. It also covers static analysis and type systems, which no current top-level branch touches, without duplicating the reproducibility-decay or dependency-reachability branches: those study environment drift and alert triage, not defect-prevention efficacy.

Highest-priority active modules right now

Title	Depth	Priority	Statement	Rationale
Produce rigorous, reproducible, paper-ready computer-science research for reputable IT conferences. Identify important problems in software engineering, computer systems, cybersecurity, data and machine-learning systems, networking, human-computer interaction, and developer tooling; build and evalua	0	1.00	Produce rigorous, reproducible, paper-ready computer-science research for reputable IT conferences. Identify important problems in software engineering, computer systems, cybersecurity, data and machine-learning systems, networking, human-computer interaction, and developer tooling; build and evaluate novel methods against credible baselines; preserve code, data provenance, tests, negative results, and result artifacts; and make only claims supported by repeatable evidence, while respecting truth, safety, legality, privacy, resource limits, and owner control.	-
Publication pivot: Across blinded unique faults and replayed real regressions on at least three public models including a branching network, does decision-aware marginal causal patching achieve higher top-1 localization than CRADLE-style anomaly propagation and activation first-divergence at the sam	2	1.00	Across blinded unique faults and replayed real regressions on at least three public models including a branching network, does decision-aware marginal causal patching achieve higher top-1 localization than CRADLE-style anomaly propagation and activation first-divergence at the same held-out false-positive rate, under a prospectively registered paired analysis?	Continue the research but stop expanding artifact-management machinery. The portfolio has crossed from concept to a genuine controlled prototype, yet its headline statistic is undermined by pseudoreplication and invalid FPR matching, and its 100% result is obtained on a benchmark structurally aligned with the method. The next cycle should spend nearly all effort on a prospectively frozen, blinded, multi-model evaluation with credible localization baselines and real regressions. If the advantage disappears on branching graphs or against CRADLE-style propagation, narrow or abandon the method claim rather than adding more synthetic families. Forced pivot: 3+ prior 'continue' verdicts did not close this gap.
Concluding deliverable: Design metamorphic testing rules for numerical drift in ML inference	6	1.00	Write a final report or paper synthesizing what was learned and built under 'Design metamorphic testing rules for numerical drift in ML inference' (Formulate metamorphic relations that can detect numerical drift in ML inference outputs when inputs are transformed in ways that should not affect the model's behavior, such as input normalization or permutation.). This branch is being concluded: codex-cli triage: This subbranch has produced a bounded, reproducible negative-result paper that publication review recommends submitting without further scope expansion; the stronger historical-defect question is already isolated in module 9db37cbb-7e94-4280-8e22-3cb0dcea4265. This is its last task - do not propose further work here, just document what came of it.	codex-cli triage: This subbranch has produced a bounded, reproducible negative-result paper that publication review recommends submitting without further scope expansion; the stronger historical-defect question is already isolated in module 9db37cbb-7e94-4280-8e22-3cb0dcea4265.
Do gradual type annotations prevent the bugs they are claimed to prevent?	1	1.00	Mine public Python repositories for commits that add type annotations and for subsequent bug-fixing commits, reconstruct each pre-fix tree, and measure what fraction of historically observed defects a static checker would have caught at the annotation level actually present, against unannotated and fully-annotated counterfactuals; report where annotations demonstrably do not help as a negative result.	Gradual typing is advocated on the premise that annotations catch real defects, but the empirical evidence is largely correlational or survey-based rather than a counterfactual test against the actual defect history of the annotated code. The design is fully reproducible from public data with local compute and a checker BrainOS already has, needs no gated resources or human subjects, and has a credible baseline (checking the same commits without annotations) and a genuine chance of a publishable negative result. It also covers static analysis and type systems, which no current top-level branch touches, without duplicating the reproducibility-decay or dependency-reachability branches: those study environment drift and alert triage, not defect-prevention efficacy.
Reproducibility decay of public research artifacts under dependency resolution drift	1	1.00	Automatically re-execute public research code artifacts under pinned, partially pinned, and freely resolved dependency sets, measure how often and how quickly results change or execution breaks, and identify which declaration practices predict survival, against a re-run-as-published baseline.	Artifact evaluation asks whether code runs today, not whether it will still run once resolvers pick newer versions, and the field lacks a controlled measurement of how fast that decays or which pinning practices actually help. The design is a genuine controlled experiment with a clean independent variable (resolution strategy) and outcomes that separate hard failure from silent numerical divergence — the latter being the more interesting result. It is entirely feasible locally: public repositories, PyPI, containerised re-execution, no gated data or hardware. It also produces a reusable harness and a negative-result-friendly finding, which is exactly what the publication feedback has been faulting this portfolio for missing.
Causality-guided reduction and localization of compiler miscompilations	1	0.98	Develop a trace- and semantics-aware method that jointly localizes and minimizes confirmed compiler miscompilations while excluding undefined-behavior explanations; evaluate on public LLVM, GCC, and Rust regression corpora against C-Reduce, Perses, differential testing, and compiler-pass bisection.	Compiler miscompilations are high-impact systems failures, and existing reducers optimize testcase size without necessarily preserving diagnostic causal structure or efficiently identifying the responsible optimization pass. Public bug corpora and open toolchains permit reproducible evaluation.
Differential detection and repair of accessibility regressions in evolving web interfaces	1	0.96	Develop a browser-accessibility-tree and interaction-trace method that localizes accessibility regressions introduced by public web-application commits and synthesizes constrained repairs; evaluate on replayed repository histories against axe-core rule reports, DOM differencing, and generic program-repair baselines, measuring confirmed regression recall, localization rank, repair validity, and visual/behavioral preservation.	The portfolio lacks an active HCI research program. Public repository histories and automated browser instrumentation permit reproducible evaluation without recruiting users, while accessibility regressions are important and current rule-based tools provide credible baselines but weak causal localization and repair.
Lineage-aware diagnosis and repair of schema-evolution failures in data pipelines	1	0.77	Build a provenance-aware method that identifies the minimal upstream schema change responsible for a broken analytical pipeline and synthesizes constrained compatibility repairs; evaluate on replayed public dbt, Airflow, and Spark repository histories against schema-diff, blame, rule-based adapter, and unconstrained repair baselines.	Schema evolution causes costly, recurring data-system failures, but ordinary diffs do not connect an upstream change to downstream semantic breakage. Historical public repositories provide natural regressions, temporal splits, and reproducible ground truth without proprietary data.
Baseline comparator: Standard rerun-based flakiness detection	5	0.68	Implement a baseline system that simply runs each test suite N times (e.g., 10) in an identical environment and flags tests with inconsistent exit codes or output diffs. Compare the 'causal localization' method's results against this naive baseline to quantify added value.	Scientific claims require comparison against credible baselines. The simplest approach is repeated execution. This module establishes the control condition for the experiment, allowing us to measure whether causal localization provides better insight or efficiency than brute-force rerunning.
Reproducibility Artifact: Dockerized Build Environment for Public Repositories	3	0.66	Create a Makefile-driven pipeline that clones 5 specific GitHub repositories known for flaky tests, installs their dependencies via pinned requirements.txt/setup.py, and runs them in isolated Docker containers. Output the raw logs and environment hashes.	Addresses the 'Reproducibility Package' parent goal directly. This is the data acquisition layer. Without a standardized way to fetch and run these builds, all subsequent analysis is unreplicable. It creates the dataset for the study.
Structured log diffing and semantic failure classification	5	0.66	Develop a parser that takes stdout/stderr from two runs (pass/fail) and generates a semantic diff. Classify failures into categories: Timeout, AssertionError, NullPointerException, StackOverflow, etc., using regex patterns or lightweight LLM-based classification if available. Output the failure type alongside the environment hash.	Raw logs are unstructured noise for statistical analysis. Categorizing failures allows us to ask finer-grained research questions (e.g., 'Are timeouts more sensitive to CPU affinity than assertions?'). This adds semantic depth to the empirical data.
Define contamination taxonomy and evaluation protocol	2	0.65	Formalize definitions of direct, indirect, and solution contamination for coding-agent benchmarks and specify measurable criteria for time-sliced auditing.	Shared definitions are needed before detection methods can be compared and claims can be falsified.
Negative result reporting: Analysis of non-flaky tests	5	0.65	Run the perturbation framework on a control set of known stable tests. Document cases where perturbations failed to induce flakiness or where causal localization failed to find a minimal cause. Analyze why (e.g., insufficient perturbation magnitude, unrelated environmental factors).	Publishing only positive results biases the scientific record. This module explicitly seeks and documents negative results and limitations, increasing the credibility and rigor of the final paper. It answers 'when does this method not work?'
Reproducibility Package: Containerized Experiment Runner and Data Schema	2	0.65	Create a Docker-based environment that encapsulates the corpus, instrumentation code, perturbation framework, and analysis scripts. Include a Makefile or script to re-run all experiments from scratch, producing identical statistical outputs. Document all random seeds and version pins.	Prior rejections heavily emphasized 'reproducibility' and 'fabricated content'. This package is the definitive proof of reproducibility. It ensures another researcher can rerun the study and verify claims, directly addressing portfolio constraints.
Exact-match provenance extraction	3	0.64	Build a deterministic exact-match detector that maps each benchmark task and reference solution to public repository snapshots, records first-seen timestamps, and emits per-task exact-match indicators with versioned provenance.	Exact matches provide the strongest contamination signal and need reliable timestamps to separate pre-release from post-release leakage.

Publication funnel

This is the scientific throughput path, including work that is not ready to become prose yet. An active publication gap is currently receiving focused problem-tree priority; a deferred gap is waiting for one of the bounded focus slots.

Current mission areas	Eligible for preflight	Publication-ready papers	Hosted calls today
55	1	1	claude: 7, codex: 2

Publication runs: not_ready 167 · rendered 2 · retryable_error 107 · quality_rejected 11

Gap records (including resolved history): active 1 · deferred 3 · resolved 142

Current readiness bottlenecks

pivot · deferred2026-08-30T09:40:08.597890+00:00

Submit this as a narrow empirical negative-result paper. Do not delay submission for the stronger historical-defect study, and do not broaden the present claim to transformers, GPUs, production workloads, or runtime-defect recall. The next research cycle should use the reproduced issues to test the alignment diagnostic prospectively. Forced pivot: 3+ prior 'continue' verdicts did not close this gap.

draft body is ~4572 words; required range is 7000-10000

continue · deferred2026-08-29T11:43:38.071646+00:00

Collect the missing source material before reassessing merit.

only 0 usable sources; need 8

pivot · active2026-08-27T21:04:39.132692+00:00

Continue the research but stop expanding artifact-management machinery. The portfolio has crossed from concept to a genuine controlled prototype, yet its headline statistic is undermined by pseudoreplication and invalid FPR matching, and its 100% result is obtained on a benchmark structurally aligned with the method. The next cycle should spend nearly all effort on a prospectively frozen, blinded, multi-model evaluation with credible localization baselines and real regressions. If the advantage disappears on branching graphs or against CRADLE-style propagation, narrow or abandon the method claim rather than adding more synthetic families. Forced pivot: 3+ prior 'continue' verdicts did not close this gap.

Replace the invalid FPR analysis with genuinely held-out calibration and evaluation. The current code hard-codes zero FPR, while the repository's own held-out audit reports a 10.7% mean and severe fold instability.
Run a prospectively frozen, powered analysis using unique faults as experimental units. The current protocol artifact postdates the raw experiments, and the combined summary double-counts runtime executions.
Demonstrate generalization beyond a benchmark-aligned linear chain by evaluating multiple public models, including at least one branching architecture, randomized layers, multiple severities, and blinded evaluation.
Add real-defect evidence by replaying independently verified historical runtime/compiler regressions or a justified external mutant corpus. The selected natural version pair shows no measurable drift, so all positive evidence is synthetic.

pivot · deferred2026-08-24T04:03:03.845608+00:00

The narrower, falsifiable question (calibrated fixed atol/rtol baseline vs. decision-aware metamorphic localization across two pinned ONNX Runtime versions) is a real, distinct research question with genuine partial infrastructure — but this cycle repeated the scaffolding-crowds-out-the-experiment pattern at a smaller scale: ~150 granular checksum/inventory/validation/packaging sub-tasks were produced, while the one substantive result sits on a compromised base. An independent sandboxed rerun of four of the project's own verification scripts failed on all four (permission error, missing numpy in the 'pinned' venv, and two missing manifest files) — the reproducibility bundle does not actually work cold, contradicting the artifact-rerun requirement. The team's own leakage analysis found the calibrated baseline's true held-out FPR is 10.7%, not the required <=5%, and explicitly flagged this as invalidating the matched-FPR claims — good, honest reporting, but it means the headline comparison built on top of it is not yet valid. Worst of all, the file claimed to be the pinned ResNet-18 model was found to actually be an HTML bot-wall page; most real experiments ran on a substituted MNIST CNN without the research question being updated to match. Do not approve this for tier-2 submission. Do not schedule more validation/checksum/packaging jobs. The next job(s) must: fix the model provenance (real ResNet-18 or an honestly reframed substitute), fix calibration leakage with a genuinely disjoint split, rebuild the repro bundle and verify it cold in a fresh sandbox, and then run exactly one clean end-to-end comparison — reporting the result honestly even if it is negative. Forced pivot: 3+ prior 'continue' verdicts did not close this gap.

Independently-verified reproducibility failure: 4/4 sampled cold reruns of the project's own verification scripts failed in an isolated sandbox (permission error, missing numpy in the 'pinned' venv, two missing manifest files), directly contradicting every claim of exact, working rerun commands
Admitted calibration leakage: the team's own 5-fold cross-validation reports a held-out FPR of 10.7%, not the required <=5%, which the team itself states invalidates every downstream matched-FPR localization number
Model integrity failure: a file named resnet18.onnx was found to be an HTML bot-wall page rather than a real model; most executed experiments instead ran on an unannounced substitute (a small MNIST CNN), which is not reflected in the stated research question or paper framing
The one completed head-to-head candidate-vs-baseline comparison is built on top of the leaky calibration and the substituted/unverified model, so its statistical results (Wilson CIs, McNemar, Cohen's h) cannot currently be trusted as evidence for the thesis

Papers

When a current research area has a real contribution, credible baselines, measured results, reproducible artifacts, and sufficient external evidence, BrainOS drafts and independently reviews a conference-structured paper. Citations are checked against gathered evidence, never invented. Only manuscripts that pass the scientific and rendered-PDF gates appear here.

Design metamorphic testing rules for numerical drift in ML inference

When Metamorphic Contrast Hurts: A Negative Result for Rotation-Conditioned Detection of Injected Neural-Network Drift Design metamorphic testing rules for numerical drift in ML inference · rendered · 2026-08-28T19:47:57.300066+00:00

Machine-learning models are rarely deployed once and left alone. The engines that run them keep changing underneath: inference software gets upgraded for speed, optimizers get switched on, numerical kernels get swapped out, all while the model file itself stays the same. These upgrades almost never crash. Instead, they can silently nudge a handful of output values just enough to flip a decision that was sitting near a boundary -- a quality-control camera on a factory line that stops flagging a scratched part, or a hospital triage tool that quietly drifts on exactly the scans it was already least confident about. Because there is no independent ground truth for what a model's output "should" be after such an upgrade, testers reach for an oracle that does not need one: run the model on an input and again on a lightly transformed copy of it (a 10-degree image rotation, which should not change what the image shows), and compare how much the two software versions disagree on each. It is tempting to subtract one disagreement from the other, on the reasoning that whatever is common to both is background noise and whatever differs is the real signal. We show this reasoning can backfire. Across three public CIFAR-100 classifiers, two pinned ONNX Runtime configurations, and 54 controlled weight-perturbation faults (16,200 faulty and 900 clean held-out cases), the subtraction-based detector caught 75.48% of injected faults versus 82.75% for simply comparing outputs on the unrotated image alone, at the same 0.67% false-alarm rate. A 5,000-replicate bootstrap that redoes the calibration itself, not just the test split, keeps this gap negative throughout (95% interval -8.80 to -3.81 points), and the same pattern holds, more sharply, for pinpointing which layer was perturbed. The mechanism is intuitive in hindsight: a persistent fault shifts behavior similarly on both the original and the rotated input, so subtracting the two disagreements discards exactly the evidence a detector needs to notice anything is wrong. For anyone building automated tests for this kind of silent drift, the practical lesson is concrete: a two-input comparison is not automatically better than a one-input one, and it must be checked against the plain baseline at a matched false-alarm rate before it is trusted to catch the next drifting camera or drifting triage tool. The finding is scoped to the tested substrate of injected weight drift, not a verdict on metamorphic testing as a strategy.

Download PDF · Download LaTeX source

Recent coding jobs

Each job is one implementation attempt, run in its own disposable workspace. "Why" is the specific decomposition step that led to it — the mission itself is shown once, above, not repeated per job (or, for owner-submitted work with no tree module, this names who asked and what they asked for) — nothing here is invented, it's what the planner actually stored at that step.

complete requested by program-attachment:improvement-cycle-3994cc67 · attempt 0 · finished 2026-08-28T02:19:08.466923+00:00

Request: Close the Program-Attachment Bottleneck for the Vacuous-Tests Branch: implement and evaluate an assertion-level vacuity detector against a hand-labelled gold set and the assertion-count baseline, then attach the result to the research area with ProgramAttachmentService.

Why: “Measuring and detecting vacuous tests in agent-written code” — The area-chair feedback states the tests in this portfolio 'assert hardcoded constants are present and files exist', and the workspaces bear that out directly: suites that assert a hardcoded table has the columns it was written with, or that a file exists after the test itself created it. Coverage does not catch this and neither did the automated reviews, so the internal quality signal is actively misleading — which is a real, general software-engineering problem now that agent-written suites are widespread. The corpus is already on disk (roughly a hundred workspaces of genuine agent-written tests), mutation testing gives an objective gold label without human raters, and the evaluation is a controlled comparison against established baselines. Contribution is a measured prevalence result, a labelled corpus, and a detector; a negative result (coverage predicts mutation score well after all) is publishable too. → “Assertion-level vacuous-test measurement program” — Implement an AST-based detector and a naive assertion-count baseline; evaluate both against a 41-row hand-labelled gold set with a deterministic harness; retain exact result artifacts and a documented rubric.

Workspace: /opt/brain/workspaces/d9d32e77-5b1a-4c19-88d0-ef65e0fff5f0/vacuous_tests_program

cancelled requested by improvement-cycle:c70648ab-19fe-4258-b4a7-9a3136b1604e · attempt 0 · finished 2026-08-28T08:56:45.631287+00:00

Request: Complete this reviewed, bounded mission-improvement task: Confirm the JSON-retry backlog recovery actually clears the 5 stuck coding_jobs coding_jobs d9d32e77 (queued since 08-11), ab2e1475 (since 08-16), 7e285265 (since 08-18), plus 2 more queued since 08-27, all carry the exact pre-a2f5e32 'all local reviewers failed'/json_invalid signature the newly-merged recover_jobs_blocked_by_fixed_review_

Why: Directly requested by improvement-cycle:c70648ab-19fe-4258-b4a7-9a3136b1604e: "Complete this reviewed, bounded mission-improvement task: Confirm the JSON-retry backlog recovery actually clears the 5 stuck coding_jobs coding_jobs d9d32e77 (queued since 08-11), ab2e1475 (since 08-16), 7e285265 (since 08-18), plus 2 more queued since 08-27, all carry the exact pre-a2f5e32 'all local reviewers failed'/json_invalid signature the newly-merged recover_jobs_blocked_by_fixed_review_"

Last error: Cancelled during owner-requested BrainOS cleanup: diagnostic or verification work is obsolete because its fix/result is already present.

Workspace: /opt/brain/workspaces/36a2a24f-9fa0-4c12-aae3-797185b76197

cancelled requested by improvement-cycle:c70648ab-19fe-4258-b4a7-9a3136b1604e · attempt 0 · finished 2026-08-28T08:56:45.631287+00:00

Request: Complete this reviewed, bounded mission-improvement task: Verify the review-timer flock fix holds across a real concurrent firing The single-flight fix (scripts/run-brain.sh, commit 2a6ac6a) was verified with a manual flock acquire/contend test but not yet observed preventing an actual concurrent 03:30/15:30 Europe/Bucharest timer firing, since only one review ran during this pass. Check audit_ev

Why: Directly requested by improvement-cycle:c70648ab-19fe-4258-b4a7-9a3136b1604e: "Complete this reviewed, bounded mission-improvement task: Verify the review-timer flock fix holds across a real concurrent firing The single-flight fix (scripts/run-brain.sh, commit 2a6ac6a) was verified with a manual flock acquire/contend test but not yet observed preventing an actual concurrent 03:30/15:30 Europe/Bucharest timer firing, since only one review ran during this pass. Check audit_ev"

Last error: Cancelled during owner-requested BrainOS cleanup: diagnostic or verification work is obsolete because its fix/result is already present.

Workspace: /opt/brain/workspaces/43ac16cf-a400-4a9b-ad04-4c9f11dc3650

complete requested by problem-tree:1d57043d-4b5d-49e0-91b9-a865a375aceb · attempt 1 · started 2026-08-27T15:03:27.597470+00:00 · finished 2026-08-27T15:41:22.879327+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Replicate negative natural drift finding Run decision-aware localization on natural version drift without faults and document negative result. Rationale: Demonstrates specificity and reduces false claims. Work only in the assigned workspace and build on its existing files. Produce a documented, testable,

Why: “Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Replicate negative natural drift finding” — Demonstrates specificity and reduces false claims.

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

cancelled requested by improvement-cycle:ac463bac-de86-4d4b-bd4e-1a95a7a30a3e · attempt 1 · started 2026-08-28T03:04:26.172022+00:00 · finished 2026-08-28T08:56:45.631287+00:00

Request: Complete this reviewed, bounded mission-improvement task: Reapply the sleep-review-cycle single-flight guard once uncontended Reapply the flock -n -E 99 /tmp/brainos-sleep-review-cycle.lock wrapper to scripts/run-brain.sh's sleep-review-cycle dispatch (full patch saved in Claude memory 2026-08-27-review-timers-collide-single-flight-fix-blocked.md), then run the quality gate and commit. Only proce

Why: Directly requested by improvement-cycle:ac463bac-de86-4d4b-bd4e-1a95a7a30a3e: "Complete this reviewed, bounded mission-improvement task: Reapply the sleep-review-cycle single-flight guard once uncontended Reapply the flock -n -E 99 /tmp/brainos-sleep-review-cycle.lock wrapper to scripts/run-brain.sh's sleep-review-cycle dispatch (full patch saved in Claude memory 2026-08-27-review-timers-collide-single-flight-fix-blocked.md), then run the quality gate and commit. Only proce"

Last error: Cancelled during owner-requested BrainOS cleanup: diagnostic or verification work is obsolete because its fix/result is already present.

Workspace: /opt/brain/workspaces/896bc9a6-d350-4b26-af08-71b389f88b1d

complete requested by problem-tree:6e1fbf10-0658-4f6c-819a-728cd29360ea · attempt 1 · started 2026-08-27T14:23:33.479410+00:00 · finished 2026-08-27T15:03:22.998078+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Implement decision-aware metamorphic oracle Build a metamorphic oracle that flags decision changes under input perturbations and weights layer contributions by decision impact. Rationale: Core method contribution for localization. Work only in the assigned workspace and build on its existing files. Produ

Why: “Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Implement decision-aware metamorphic oracle” — Core method contribution for localization.

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

complete requested by problem-tree:53d4fa38-99aa-4ea2-8c86-45417c735f2b · attempt 1 · started 2026-08-27T12:59:41.022473+00:00 · finished 2026-08-27T14:23:33.314115+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Construct single-layer fault injection benchmark Generate at least 30 single-layer fault injections across three fault families with matched negative controls for MNIST CNN. Rationale: Provides powered sample for statistical test. Work only in the assigned workspace and build on its existing files. Produ

Why: “Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Construct single-layer fault injection benchmark” — Provides powered sample for statistical test.

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

complete requested by problem-tree:aad57050-b748-4d65-8be2-b574a066bc37 · attempt 2 · started 2026-08-27T12:00:12.434971+00:00 · finished 2026-08-27T12:44:46.479304+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Run pre-registered localization experiment Execute decision-aware vs fixed tolerance localization on the fault injection benchmark with pre-registered hypothesis and collect top-1 accuracy. Rationale: Produces primary empirical evidence. Work only in the assigned workspace and build on its existing files

Why: “Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Run pre-registered localization experiment” — Produces primary empirical evidence.

Last error: Codex OSS could not produce an independently verified implementation Steps taken: ValidationError: 1 validation error for CodingResult limitations Input should be a valid array [type=list_type, input_value='Single model architectur...ld version regressions.', input_type=str] For further information visit https://errors.pydantic.dev/2.13/v/list_type

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

cancelled requested by problem-tree:4148fcf4-6c54-4461-98c9-c069885c9797 · attempt 3 · started 2026-08-27T12:44:50.788515+00:00 · finished 2026-08-27T21:04:38.978472+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Build pinned ONNX Runtime harness Create a reproducible harness that runs MNIST CNN and ResNet-18 on ONNX Runtime 1.20.0 vs 1.28.0 CPU with deterministic seeds and logs outputs. Rationale: Enables controlled version-drift experiments. Work only in the assigned workspace and build on its existing files. P

Why: “Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Build pinned ONNX Runtime harness” — Enables controlled version-drift experiments.

Last error: replaced by a newer pivot verdict on the focused work module

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

complete requested by problem-tree:167707b2-5898-4771-bea4-6f2438add7b1 · attempt 1 · started 2026-08-27T10:39:56.295714+00:00 · finished 2026-08-27T11:17:32.309534+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Implement fixed atol/rtol differential baseline Implement a reproducible fixed absolute/relative tolerance differential tester for ONNX Runtime outputs and log false positive rate. Rationale: Provides credible baseline for localization comparison. Work only in the assigned workspace and build on its exis

Why: “Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Implement fixed atol/rtol differential baseline” — Provides credible baseline for localization comparison.

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

complete requested by problem-tree:8cc4a970-0c78-4c26-9c52-48172e801aae · attempt 1 · started 2026-08-27T09:43:54.720674+00:00 · finished 2026-08-27T10:10:07.885091+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Package reproducible artifacts Create manifest with SHA256 checksums, verifier script, and provenance log for all code and data. Rationale: Ensures paper-ready reproducibility. Work only in the assigned workspace and build on its existing files. Produce a documented, testable, reusable research artifact.

Why: “Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Package reproducible artifacts” — Ensures paper-ready reproducibility.

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

cancelled requested by problem-tree:eb13fbb1-ece2-415b-be59-c7fbb3848b36 · attempt 2 · started 2026-08-27T10:10:12.317322+00:00 · finished 2026-08-27T10:54:01.988636+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Quarantine fake artifacts list Generate a quarantine manifest listing known fake artifacts removed and reason for removal. Rationale: Provides auditable record of contamination remediation. Work only in the assigned workspace and build on its existing files. Produce a documented, testable, reusable resea

Why: “Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Quarantine fake artifacts list” — Provides auditable record of contamination remediation.

Last error: queued execution retired because its problem-tree module is abandoned

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

cancelled requested by problem-tree:852b683b-2b81-4f6f-bb88-b19052e66cb0 · attempt 1 · started 2026-08-27T07:22:55.102418+00:00 · finished 2026-08-27T10:54:01.981834+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Cross-check file metadata vs manifest Validate size bytes and modification time for each verified file against manifest entries. Rationale: Detects drift between filesystem and recorded metadata. Work only in the assigned workspace and build on its existing files. Produce a documented, testable, reusable

Why: “Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Cross-check file metadata vs manifest” — Detects drift between filesystem and recorded metadata.

Last error: queued execution retired because its problem-tree module is abandoned

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

complete requested by problem-tree:f1722bc6-7341-40dc-bd9d-18e8e2a64ab8 · attempt 2 · started 2026-08-27T07:51:06.146335+00:00 · finished 2026-08-27T08:38:21.884190+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Atomic manifest rewrite with checksums Write manifest.json atomically with updated digests, sizes and mtimes. Rationale: Atomic write avoids partial updates that corrupt reproducibility records. Work only in the assigned workspace and build on its existing files. Produce a documented, testable, reusable

Why: “Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Atomic manifest rewrite with checksums” — Atomic write avoids partial updates that corrupt reproducibility records.

Last error: RuntimeError: all local reviewers failed: muse-glimmer-coding:30b-q4_K_M: HostAgentTimeout: Codex agent exceeded its 600s budget

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

complete requested by problem-tree:f72ce076-076f-4179-94fe-305bdfaaac52 · attempt 1 · started 2026-08-27T05:52:24.542549+00:00 · finished 2026-08-27T06:28:48.949193+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Regenerate bundle verifier script Create or update a small verifier script that checks file existence, sizes and SHA256 against manifest. Rationale: Provides reusable evidence that bundle is intact for reviewers. Work only in the assigned workspace and build on its existing files. Produce a documented, t

Why: “Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Regenerate bundle verifier script” — Provides reusable evidence that bundle is intact for reviewers.

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

complete requested by problem-tree:f19bae63-42d3-4e5e-b162-4f60d7218449 · attempt 1 · started 2026-08-27T05:21:41.156272+00:00 · finished 2026-08-27T05:52:22.558062+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Validate manifest schema and existence Check outputs/bundles/v1_2026-08-18/manifest.json exists and conforms to expected schema with file entries. Rationale: Missing or malformed manifest breaks reproducibility bundle verification. Work only in the assigned workspace and build on its existing files. Prod

Why: “Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Validate manifest schema and existence” — Missing or malformed manifest breaks reproducibility bundle verification.

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

complete requested by problem-tree:d5d58318-9f63-430b-abdd-1d347b3b9588 · attempt 1 · started 2026-08-27T05:08:40.697295+00:00 · finished 2026-08-27T05:21:40.950305+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Run independent integrity verification test Execute verifier script on the regenerated bundle and capture exit code and output. Rationale: Confirms the remediation produces a verifiable bundle without manual inspection. Work only in the assigned workspace and build on its existing files. Produce a docume

Why: “Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Run independent integrity verification test” — Confirms the remediation produces a verifiable bundle without manual inspection.

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

cancelled requested by problem-tree:e4334540-dd55-408b-ac09-efbc6f933032 · attempt 1 · started 2026-08-27T04:40:10.549342+00:00 · finished 2026-08-27T10:54:01.969532+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Purge synthetic per_trial_results.csv Delete synthetic per_trial_results.csv and verify removal to remove contaminated evaluation data. Rationale: Contaminated CSV undermines benchmark integrity and could be re-ingested by automation. Work only in the assigned workspace and build on its existing files. P

Why: “Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Purge synthetic per_trial_results.csv” — Contaminated CSV undermines benchmark integrity and could be re-ingested by automation.

Last error: queued execution retired because its problem-tree module is abandoned

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

complete requested by problem-tree:abc3f486-cda7-4a36-bdf3-8d27bc0922ce · attempt 1 · started 2026-08-27T04:03:39.248219+00:00 · finished 2026-08-27T04:40:10.397888+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Compute SHA256 for bundle files Compute SHA256 checksums for all verified bundle files under outputs/bundles/v1_2026-08-18. Rationale: Checksums enable integrity verification and detect silent corruption. Work only in the assigned workspace and build on its existing files. Produce a documented, testable,

Why: “Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Compute SHA256 for bundle files” — Checksums enable integrity verification and detect silent corruption.

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

complete requested by problem-tree:5c6b2723-9d1e-4711-bff6-97156cb707f5 · attempt 1 · started 2026-08-27T03:48:42.154541+00:00 · finished 2026-08-27T04:03:39.111206+00:00

Request: Complete this bounded scientific objective from BrainOS's computer-science research program: Verify deletion of models/resnet18.onnx Remove models/resnet18.onnx from workspace and confirm absence via filesystem stat and existence checks. Rationale: Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. Work only in the assigned workspace and build on

Why: “Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Verify deletion of models/resnet18.onnx” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification.

Workspace: /opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e

Workspaces

A workspace is a dedicated, reversible git working tree under /opt/brain/workspaces/<job-id> on the server, one per coding job above — this is where the "why" above actually gets implemented. Listed here is what currently sits at the top level of each one; look inside further over SSH at the path shown.

/opt/brain/workspaces/d9d32e77-5b1a-4c19-88d0-ef65e0fff5f0/vacuous_tests_program exists

“Measuring and detecting vacuous tests in agent-written code” — The area-chair feedback states the tests in this portfolio 'assert hardcoded constants are present and files exist', and the workspaces bear that out directly: suites that assert a hardcoded table has the columns it was written with, or that a file exists after the test itself created it. Coverage does not catch this and neither did the automated reviews, so the internal quality signal is actively misleading — which is a real, general software-engineering problem now that agent-written suites are widespread. The corpus is already on disk (roughly a hundred workspaces of genuine agent-written tests), mutation testing gives an objective gold label without human raters, and the evaluation is a controlled comparison against established baselines. Contribution is a measured prevalence result, a labelled corpus, and a detector; a negative result (coverage predicts mutation score well after all) is publishable too. → “Assertion-level vacuous-test measurement program” — Implement an AST-based detector and a naive assertion-count baseline; evaluate both against a 41-row hand-labelled gold set with a deterministic harness; retain exact result artifacts and a documented rubric.

README.md, baseline.py, detector.py, evaluate.py, gold.csv, gold_set.py, results.json, rubric.md, test_detector.py

/opt/brain/workspaces/36a2a24f-9fa0-4c12-aae3-797185b76197 exists

Directly requested by improvement-cycle:c70648ab-19fe-4258-b4a7-9a3136b1604e: "Complete this reviewed, bounded mission-improvement task: Confirm the JSON-retry backlog recovery actually clears the 5 stuck coding_jobs coding_jobs d9d32e77 (queued since 08-11), ab2e1475 (since 08-16), 7e285265 (since 08-18), plus 2 more queued since 08-27, all carry the exact pre-a2f5e32 'all local reviewers failed'/json_invalid signature the newly-merged recover_jobs_blocked_by_fixed_review_"

BRIEFING.json

/opt/brain/workspaces/43ac16cf-a400-4a9b-ad04-4c9f11dc3650 exists

Directly requested by improvement-cycle:c70648ab-19fe-4258-b4a7-9a3136b1604e: "Complete this reviewed, bounded mission-improvement task: Verify the review-timer flock fix holds across a real concurrent firing The single-flight fix (scripts/run-brain.sh, commit 2a6ac6a) was verified with a manual flock acquire/contend test but not yet observed preventing an actual concurrent 03:30/15:30 Europe/Bucharest timer firing, since only one review ran during this pass. Check audit_ev"

BRIEFING.json

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Replicate negative natural drift finding” — Demonstrates specificity and reduces false claims.

.brainos-codex-complete2, .brainos-codex/, .brainos-context/, .brainos-manifest.json, .brainos/, .git/, .gitignore, .venvs/, PROJECT_STATUS.md, README_metamorphic_oracle.md, README_onnx_activation_hook.md, README_oracle.md, README_oracle_harness.md, README_quarantine_manifest.md, README_ranking.md, VERIFICATION_resnet18_deletion.md, activation_quantizer_injector.py, aggregate_baseline_metrics.py, aggregate_baseline_metrics.py.bak, apply_calibrated_thresholds_quantization.py, apply_calibrated_thresholds_quantization_realtime.py, apply_fixed_thresholds_additive.py, assess_calibration_leakage.py, atomic_write.py, breakdown_by_fault_family.py, bundle_artifacts.py, bundle_artifacts.py.bak, calibrate_per_layer_atol_rtol_fpr.py, capture_activations_runtime_a.py, capture_activations_runtime_a_v2.py

/opt/brain/workspaces/896bc9a6-d350-4b26-af08-71b389f88b1d exists

Directly requested by improvement-cycle:ac463bac-de86-4d4b-bd4e-1a95a7a30a3e: "Complete this reviewed, bounded mission-improvement task: Reapply the sleep-review-cycle single-flight guard once uncontended Reapply the flock -n -E 99 /tmp/brainos-sleep-review-cycle.lock wrapper to scripts/run-brain.sh's sleep-review-cycle dispatch (full patch saved in Claude memory 2026-08-27-review-timers-collide-single-flight-fix-blocked.md), then run the quality gate and commit. Only proce"

.brainos-context/, BRIEFING.json, brainos-source/, pytest.ini

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Implement decision-aware metamorphic oracle” — Core method contribution for localization.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Construct single-layer fault injection benchmark” — Provides powered sample for statistical test.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Run pre-registered localization experiment” — Produces primary empirical evidence.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Build pinned ONNX Runtime harness” — Enables controlled version-drift experiments.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Implement fixed atol/rtol differential baseline” — Provides credible baseline for localization comparison.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge synthetic results file” — Removes fabricated experimental data that would compromise integrity of published results. → “Package reproducible artifacts” — Ensures paper-ready reproducibility.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Quarantine fake artifacts list” — Provides auditable record of contamination remediation.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Cross-check file metadata vs manifest” — Detects drift between filesystem and recorded metadata.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Atomic manifest rewrite with checksums” — Atomic write avoids partial updates that corrupt reproducibility records.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Regenerate bundle verifier script” — Provides reusable evidence that bundle is intact for reviewers.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Validate manifest schema and existence” — Missing or malformed manifest breaks reproducibility bundle verification.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Run independent integrity verification test” — Confirms the remediation produces a verifiable bundle without manual inspection.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Purge synthetic per_trial_results.csv” — Contaminated CSV undermines benchmark integrity and could be re-ingested by automation.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Compute SHA256 for bundle files” — Checksums enable integrity verification and detect silent corruption.

/opt/brain/workspaces/projects/1c692942-e01f-4f9f-a690-f07e81fb075e exists

“Purge unauthorized model file” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification. → “Verify deletion of models/resnet18.onnx” — Eliminates known fake artifact that invalidates reproducibility claims and downstream verification.

Documentation

README — what BrainOS is and how it's set up
TODO — what's built, what's known-missing, what's deliberately deferred
Architecture — the system design