PASS Is a Claim, Not an Observation: Mechanical Re-verification of 10 Self-Asserted Fix Claims in a Live Multi-Agent Audit Layer

Lei Zhenbing; Lei Chang

Home/Archive/arXiv

arXiv · 2610.35950

PASS Is a Claim, Not an Observation: Mechanical Re-verification of 10 Self-Asserted Fix Claims in a Live Multi-Agent Audit Layer

Lei Zhenbing·Lei Chang

View original source ↗Read source document ↗Export metadata ↓

Abstract

When agents claim to have fixed their own defects, how do we know the fix is real? We mechanically re-verified 10 self-asserted code-level fix claims from a privileged coordination agent (azone_god) operating in AITRAP, a live multi-agent knowledge-tree platform. These 10 claims were extracted from a corpus of 179 coordination broadcasts (28 containing structured fix assertions, 31 unique fix-ID/message pairs, 18 distinct fix identifiers). Fix claims use mixed English/Chinese formats (PASS, checkmarks, and "verification passed" declarations). Each code-level claim was checked against three criteria: (1) file modification timestamp (mtime) consistency -- 70% pass; (2) presence of the claimed defensive pattern at the specified identifier and call site -- 60% pass in the claimed file (80% anywhere in the codebase); (3) deterministic rule-based re-verification from the broadcast announcement alone -- 80% pass. The critical finding: among claims that provide sufficient information for re-verification (C3-pass), 50% fail the identifier+call-site check (C2-fail) -- the fix either does not exist at the claimed location or does not exist at all. This demonstrates that self-asserted fix claims cannot be trusted without independent mechanical verification. Our central claim: a PASS declaration is a claim requiring independent verification, not an observation of fix completion.

Explore related subjects

cs.MA cs.SE

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Lei Zhenbing, Lei Chang. 2026-10-24. PASS Is a Claim, Not an Observation: Mechanical Re-verification of 10 Self-Asserted Fix Claims in a Live Multi-Agent Audit Layer. https://arxiv.org/abs/2610.35950

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

MAS-Loop: Beyond One-Shot Orchestration towards Self-Evolving Multi-Agent Systems

Training-time multi-agent system (MAS) design seeks to learn reusable orchestration policies, yet prevailing approaches remain largely one-shot: an Orchestrator generates a system but only observes and learns from its terminal outcome. This paradigm discards execution trajectories that expose system-level failures and contain fine-grained information beyond the terminal outcome, leaving no direct basis for revising the source MAS, evaluating the updated MAS, or learning from the resulting contrast. To this end, we introduce MAS-Loop, an execution-grounded framework that closes this feedback loop by coupling an Orchestrator with a trajectory-conditioned Refiner. For each task, the Orchestrator constructs and executes a source MAS to obtain an outcome score and an intermediate trajectory. Conditioned on the trajectory, the Refiner localizes an actionable failure, formulates a trajectory-grounded diagnosis, and proposes a bounded patch to the agents, information flow, or execution configuration. The executable updated MAS is re-executed on the same task to measure its realized outcome under stochastic execution. This pair-MAS comparison provides complementary learning signals: the Orchestrator learns to generate strong systems with fewer recoverable failures, whereas the Refiner learns to expose and repair observed weaknesses, thereby driving policy-level co-evolution across training iterations. Experiments show that MAS-Loop achieves the strongest overall performance among the evaluated baselines, with consistent gains from refinement across all evaluated benchmarks. The co-evolution process also yields MAS-JudgeCorpus, a dataset of execution-grounded source-updated pairs with orchestration- and trajectory-level supervision, which can provide the community with support for MAS reward modeling and trajectory-aware investigation.

cs.MA↗

arXiv2026

When Is Fresh AI Stock Analysis Worth It? Allocating a Limited Research Budget Across Price Updates, Deep Forecast Refreshes, and Editorial Reviews of Top Stock Picks

Must long-horizon AI investment research be rewritten every time stock prices move? Using 8,120 production reports on the same 203 stocks across four iPulse AI batches, we compare repricing existing targets, guarding them against price shocks, mechanically resetting old forecast growth, and generating fresh multi-agent reports. Ten matched configurations represent seven investment frameworks. Our primary economic evaluation selects Top 10 USD portfolios using one-year forecasts. Across two unchanged-model windows, guarded reuse averages 2.25%, versus 1.66% for SPY and 4.77% for fresh reports, after modeled trading costs and reference compute charges on a USD 100,000 account. Guarded reuse beats SPY in all three short windows, including the model-change transition. Extending the first holding window to nearly six months reverses the result: guarded reuse returns -11.26%, fresh reports +9.12%, and SPY +3.46%, motivating the thesis half-life question. Supporting quarter-forecast diagnostics show that mechanically re-anchoring old growth paths closes 98% of the MAE improvement over fixed targets, yet Fresh and Reset share only 1-4 of their Top 10 stocks. At Top 30, the unchanged-model portfolio comparison instead favors guarded reuse (4.17% versus 1.88%). Endpoint and concentration tests reveal the influence of horizon choices and exceptional winners. Nine production editions show substantial between-edition turnover, while every final editorial screen retains its own incoming list. Measured token workloads connect refresh costs to shared capital and realized break-even budgets. The resulting architecture separates continuous price monitoring from episodic thesis rebuilding and targeted editorial evidence checks.

cs.MA↗

arXiv2026

AttriSwarm: Orchestrating Agent Swarms for Cross-Page Visual Attribution

Multimodal large language models (MLLMs) increasingly address knowledge-intensive tasks by integrating evidence scattered across multiple document pages into a final answer. Because such answers are synthesized from many pages, the link between the answer and its supporting visual evidence can become obscured, hindering verification and trust. Cross-page visual attribution is therefore essential: it requires identifying the supporting pages and localizing fine-grained evidence regions within them, thereby making multi-page reasoning transparent and auditable. Existing approaches either process multiple pages jointly or iteratively refine evidence within a single trajectory, risking cross-page context pollution and, for interactive methods, substantial sequential overhead. We introduce \textbf{AttriSwarm}, an agent-swarm framework that assigns these concerns to separate roles: a leader handles global cross-page reasoning, while subagents perform page-wise attribution in isolation and in parallel. To support training and evaluation, we construct \textbf{AttriDoc}, a cross-page QA dataset linking supporting pages to their annotated evidence regions, and use it for role-specific training of AttriSwarm, improving answer accuracy, page-level $F_1$, and region-level IoU over its training-free counterpart by an average of 18.94, 29.34, and 23.04 percentage points, respectively, across three open-source backbones and three datasets.

cs.MA↗