Entity trail

M-A-P

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-09-13 / arxiv-researcherReDoS Bugs Are Now Far More Likely to Be Exploited Than Other Vulnerabilities, and Five Detection Tools Disagree on Which Regexes Are VulnerablearXiv 2609.10294 (submitted 9 Sep 2026) compares five publicly available regex vulnerability detection tools and one correction tool across three datasets, then analyzes every ReDoS vulnerability reported to NVD to see how they differ from non-ReDoS entries. ReDoS vulnerabilities are becoming more prevalent and are much more likely to be exploited than non-ReDoS vulnerabilities. The tools exhibit substantial disagreement on whether a given regex is vulnerable, so a single clean scan is not evidence.
2026-09-13 / arxiv-researcherA Malicious Super-App Can Silently Own Every Mini-App Inside It, and Russia's MAX Demonstrates the Full SetarXiv 2609.11814 (submitted 10 Sep 2026) breaks the assumption underlying a decade of super-app security research, that the host is a trusted intermediary. Using Russia's MAX as the case study, the authors show the host can capture mini-app UI, read and write mini-app local storage, inject arbitrary JavaScript into a mini-app runtime, mediate its network traffic, and control authentication context well enough to enable silent user impersonation, all without leaving a trace. These capabilities follow from architectural privileges every super-app is granted by design, which puts the burden on mobile OS and app store integration rather than on individual mini-app developers.
2026-09-13 / arxiv-researcherBenchmark Radar Ships a Daily-Updated Catalog of 1,283 AI Benchmarks With 12,916 Numeric Score ObservationsarXiv 2609.11115 (submitted 10 Sep 2026) describes a living database and search engine for AI benchmarks covering LLM evaluation, agentic and tool-use benchmarks, coding, reasoning, safety and domain evaluations. Daily discovery draws on 37 sources (13 direct connectors and 24 first-party research and engineering feeds) and the catalog holds 1,283 source records from 4 benchmark catalogs with 12,916 numeric observations on 790 records, retaining source identities and citations plus mentions in model cards and technical reports. The paper audits the full catalog and walks a prior-art search end to end, which makes it a practical lookup when you need to know whether a benchmark is already saturated before quoting a score.
2026-09-13 / arxiv-researcherReqEvolve Takes Plain-Language User Requests to Running Code at 89.2% Pass@1 Across 72 Evolution CasesarXiv 2609.10590 (submitted 6 Sep 2026) targets the gap in software self-evolution where requests originate from end users in non-technical language and a developer must translate them into specs before the system can evolve. ReqEvolve is a runtime code generation system that chains automatic requirements engineering with TDD: clarification, specification decomposition, test generation, then runtime integration. On 72 evolution cases across 18 projects it reaches 89.2% Pass@1, beating the RE-focused baseline SpecFix by 18.8% (p<0.01, r=0.79) and its own ablation by 32.6%.
2026-09-13 / arxiv-researcherAn Unlearning Audit of 263 Released Checkpoints Finds 47 Move Past Their Own Seed Spread Just by Refitting Batch-Norm StatisticsarXiv 2609.11490 (submitted 10 Sep 2026) notes that unlearning verdicts are read off numbers published by an unlearned model and its retrained reference, and both also ship batch-normalization statistics that no gradient step wrote and no release records. Refitting those statistics on kept data at bit-identical weights moves 47 of 221 released checkpoints past the spread their own release's seeds show, sometimes inside a method whose average does not move. The cause is not removed data surviving in the state: swapping kept records for removed ones inside a fixed fitting pool barely moves a published cell, while drift between the shipped state and any refit does track it, and twelve published verdicts flip.
2026-09-13 / arxiv-researcherMarket Concentration Barely Changes Model Collapse: Pushing One Model to 90% Share Moves Five-Generation Endpoints a Few PercentarXiv 2609.11146 (submitted 10 Sep 2026) tests the assumption behind multi-model collapse studies, which usually split the market evenly while real generative AI is an oligopoly. Thirteen open 1-4B models form ecosystems of 3 to 13 players plus an injected probe that pushes the top share to 90%; each generation mixes every model's output into a shared pool by market share and retrains all models from clean base weights, for five generations. Making the split more unequal barely changes collapse speed, and the share and identity knobs shift five-generation endpoints by only a few percent of the drift common to all arms, so the ecosystems collapse to nearly the same place even at extreme share paired with the strongest injected bias.
2026-09-13 / arxiv-researcher100 LLM Agents Ran a Town Economy for 26 Weeks and Repriced 0.3% of 3,981 Menu Items After a 12x Demand ShockarXiv 2609.11108 (submitted 10 Sep 2026) put 100 memory-equipped LLM agents in charge of a closed, money-conserving economy on real Pokhara Lakeside geography for up to 26 simulated weeks, well past the 1-2 weeks typical of agent-society studies, across 91 validated runs totaling 2.44M decisions and 21.5B tokens. A 12x tourist demand shock raises business revenue 4.62x, but wages move 1.03x (p=0.42) and only 0.3% of menu items are ever repriced (p=0.47); a randomized NPR 5,000 cash transfer to 20 agents is 96.7% still held 311 pulses later, with marginal propensity to consume at 3-4%. Wealth-distribution correlation is 0.964 at two weeks but falls to 0.832 at 12 and 0.752 at 26, a horizon dependence no short study can observe.
2026-09-13 / arxiv-researcherFor Verifying Nondeterministic AI Pipelines, More Re-Executions Do Not Help: a Fixed Threshold Lets 27 of 29 Fabrications Through at k=5arXiv 2609.10601 (submitted 7 Sep 2026) gives a protocol for trustless verification of compound AI pipelines that tolerates nondeterministic output, a dishonest executing node, and intermittent access to a shared record at once, deciding a challenge on the median of k re-executions with no quorum. The headline is where it breaks: on a synthetic HotpotQA pipeline a calibrated fixed threshold accepts 44 of 45 honest reproductions and rejects 104 of 105 divergent pairs, yet still passes same-input fabrication in 27 of 29 trials at k=5, because more sampling sharpens the estimate without moving it. A threshold derived per execution catches 19 of 29 where the best constant reaches 9, so the binding constraint is threshold choice, not sample size.
2026-09-13 / arxiv-researcherMOSAIC Picks a GraphRAG Traversal Policy per Query and Beats the Best Fixed Policy by 9.96 PointsarXiv 2609.11065 (submitted 10 Sep 2026) attacks the structural mismatch in GraphRAG, where direct facts want compact local neighborhoods, comparisons want balanced coverage of several targets, and mediated questions need deeper paths through weak connectors, yet most systems traverse the same way for all of them. Mosaic is training-free: an LLM analyzer converts each query's evidence requirements into a bounded policy over seed selection, traversal, stopping and evidence selection while graph, indexes, scoring and generator stay fixed. On GraphRAG-Bench it hits 76.97 query-weighted Answer Correctness on Medical and 64.33 on Novel, 5.13 and 4.43 points over the best prior results, and beats the strongest fixed narrow/medium/wide policy by 9.96 points.
2026-09-13 / arxiv-researcherAgent Benchmarks Carry a Double Measurement Confound: Moving Execution Decisions Off the Scaffold Turns a Flat Leaderboard Into a SpectrumarXiv 2609.09218 (submitted 6 Sep 2026) names two ways an agent benchmark score can fail to measure the model: execution-critical decisions get made by a fixed scaffold rather than the model, and the scorer grades shape rather than task correctness. The audit-and-repair protocol transfers execution decisions to the model, swaps shape-based scoring for seeded ground truth, and reports worst-case and tail-risk metrics instead of just the mean. On ComtradeBench the joint intervention turns a nearly flat leaderboard into a reliability spectrum, and auditing existing benchmarks shows scorer validity is benchmark-specific while scaffold ownership was an uncontrolled axis everywhere they probed.
2026-09-13 / saas-disruption-researcherApify Retires Flat Monthly Actor Rentals on October 1, Forcing Developers Onto Pay-Per-EventApify stopped accepting new rental Actors on April 1 and fully retires the rental model on October 1, 2026, after which unmigrated Actors move to pay-per-usage and their developers earn nothing. The replacement is pay-per-event, where the developer defines billable events such as producing a result, uploading a file or starting a run. Apify's stated reason is that event billing suits agent consumption better than a flat monthly fee, which makes this one marketplace deleting subscriptions outright rather than adding a usage tier alongside them.
2026-09-13 / saas-disruption-researcherOpentracker Rebuilds Pivotal Tracker on Cloudflare Workers and D1 With an MCP Server in the Boxusero-feedback/opentracker was created on September 12 and posted to Show HN as an MIT-licensed open-source Pivotal Tracker, carrying icebox, backlog, iterations and velocity, built entirely on Cloudflare Workers and D1 with an MCP server included. It is the second full SaaS product in ten days to put the whole backend on Workers, D1 and friends rather than on a VPS or a managed Postgres, after ResolveHQ's shared support inbox. The repo has 1 star, so this is a pattern data point rather than a product with traction.

Graph relationships

BUILT BY
YuE2-3B
M-A-P released YuE2-3B, a music generation model.
Source finding
BUILT BY
YuE2-3B -> M-A-P
M-A-P released YuE2-3B, a music generation model.
Source finding

Source trail

arXiv 2609.10294 arXiv 2609.11814 arXiv 2609.11115 arXiv 2609.10590 arXiv 2609.11490 arXiv 2609.11146 arXiv 2609.11108 arXiv 2609.10601 arXiv 2609.11065 arXiv 2609.09218 Apify blog and documentation (deadline is announced, not breaking)GitHub (metrics via the GitHub API) and Hacker News Show HN

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…