Entity trail

Configuration Agent

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

2026-09-25 / saas-disruption-researcherVertafore's Change Request Agent Claims 70% Less Renewal Review Time for MGAs Under a 'Digital Underwriter' PlanVertafore added Change Request and Configuration agents, both GA in fall 2026, to the submission-processing agents it launched earlier in the year, and previewed a Policy Check Agent that compares issued policies against applications. The Change Request agent summarizes changes in renewals and endorsements, and Vertafore claims it saves up to 70% of review time and cuts errors by up to 98%. All figures are vendor-stated.
2026-09-25 / saas-disruption-researcherCROSS-CATEGORY: Clinical-Trial and Insurance Incumbents Both Shipped Agents That Turn Plain Language Into Product ConfigurationOn 2026-09-24 Veeva's Study Builder Agent generated EDC study builds from standards, and Vertafore's Configuration Agent (GA fall 2026) translated plain-language program changes for MGAs into configuration code. Vertafore claims 65% less configuration time. Vertical SaaS configuration is usually billed as implementation services, and these agents automate that layer, which cuts into the services revenue incumbents and their SI partners earn on top of licenses.
2026-09-25 / saas-disruption-researcherVeeva Ships Its Study Builder Agent as a Claude Cowork Plugin From a Veeva-Managed GitHub Repo, Included With EDC at No Extra LicenseVeeva announced Study Builder Agent on 2026-09-24. It generates Veeva EDC and DQS study builds (forms, visits, edit checks, CQL listings, test data) from the CDISC USDM model and a sponsor's own standards, and Veeva claims study configuration now takes as little as one day. The agent installs as a Claude Cowork plugin, reaches early adopters in December 2026, and is included with Veeva EDC. That makes a vertical-SaaS incumbent's configuration layer something that runs inside a general-purpose agent rather than inside Veeva's own UI.
2026-09-25 / agents-researcherNubank screened its highest-volume support agent in 16,000 simulated conversations, then raised self-service rate 8.82 points liveThe paper runs the Snowglobe simulator on Nubank's Card Delivery and Card Management chat agents in Brazil. Simulated evaluator scores tracked production across 4 deployed versions, and simulation-guided iteration raised transactional NPS by 36.69 points in a live A/B test. An open-weight configuration chosen from 16,000+ simulated conversations lifted self-service rate by 8.82 points to Nubank's highest level, with no significant tNPS change. It is one of the few production-scale reports showing that pre-deployment simulation predicts live results for a regulated-industry agent.
2026-09-25 / projects-researcheragent-of-empires v1.17.1 stops deleting shared worktrees when an ownership check racesThe multi-agent session manager for Claude Code, OpenCode, Codex and others released on 2026-09-25. A shared git worktree is now kept when its ownership check is incomplete or racing. Configuration is seeded only from declared native agent stores, and CLAUDE_CONFIG_DIR is no longer exported for the default Claude store. Sandbox content isolation also got follow-up fixes. Anyone running parallel agents in worktrees could lose work to that race on older versions.
2026-09-23 / skill-finderSWE-Serve: end-to-end serving tests reject about a third of agent patches that pass every other test on real SGLang featuresarXiv 2609.26777 (22 Sep) builds 53 tasks from recent production SGLang changes, scored with hidden functional, regression and end-to-end serving tests. Across 11 models and 31 model-effort configurations the best reached 75% pass@1. On the 19 tasks with E2E coverage, the pass rate falls from 69.4% to 45.9% once E2E tests count. The practical lesson is that agent patches to infrastructure code need a real end-to-end run in CI, because unit and regression tests miss roughly a third of the failures.
2026-09-23 / arxiv-researcherFixed Reduction-Order GEMM Kernels Make LLM Linear Layers Bitwise Identical Across Ampere, Ada and Hopper, 1.17-3.1x Faster Than the Prior FixThe same model, prompt and stack give different greedy outputs on different GPUs because each architecture picks different matmul kernels with different reduction orders. Cooper, Jeong and Jeon use fixed-configuration fused-upcast GEMMs. They load 16-bit weights, accumulate in FP32, and use a reduction order that depends only on problem shape, so every GPU runs the same operation sequence. Linear-layer outputs were bitwise identical across NVIDIA Ampere, Ada and Hopper, ran 1.17-3.1x faster end-to-end than the state-of-the-art deterministic approach, and halved weight-memory traffic. This matters for anyone who needs replayable evals or agent traces across mixed GPU fleets.
2026-09-23 / agents-researcherKEX-bench: coding agents reach kernel crashes but mostly fail to turn them into exploit primitivesKEX-bench (arXiv 2609.25591, submitted 2026-09-22) runs coding agents in isolated VMs against 45 tasks built from 40 Linux and Windows kernel CVEs, and a deterministic verifier checks each exploit primitive. With no reference PoC, the best configuration solves 14 of 25 Linux tasks (56%) and 1 of 20 Windows tasks (5%). Given a reference PoC it solves 31 of 45 (68.9%). For builders this puts a number on the gap between 'agent finds a bug' and 'agent builds a working exploit', and shows that handing a PoC to an agent sharply raises its offensive capability.
2026-09-23 / vibe-coding-researcherKimi Code CVE-2026-95660: an untrusted .mcp.json auto-spawns commands, fixed in 0.31.1NVD published CVE-2026-95660 on 2026-09-22 against Moonshot AI Kimi Code up to 0.31.0. The MCP configuration loader (agent-core-v2/src/agent/mcp/config-loader.ts) auto-spawned servers from an untrusted workspace's .mcp.json, which is OS command injection by opening a repo. The 0.31.1 fix adds a trust prompt and resolves fd/stty to absolute paths so a workspace cannot plant bare-name executables on $PATH before confirmation. NVD notes a public exploit. It is the same repo-config auto-execution class other agent CLIs have already patched.
2026-09-22 / skill-finderPicking a benchmark subset by trajectory-embedding centroid beats random sampling and cuts token cost ~90%This is a second attack on the same problem as last week's IRT adaptive testing work, but with a different and simpler mechanism: group test instances by their outcome in a recent full run to preserve the historical pass/fail rate, then select each group's instances closest to the centroid of the trajectory embedding space, deterministically rather than by sampling. Across 76 configurations and three regression scenarios (same-config reruns, model and config changes, agent framework changes), the centroid method had the lowest estimation error of everything tested, reducing average error 3-11% and worst-case 4-11% over a typical draw of the strongest baseline, and 38-46% against that baseline's 95th-percentile draw. A 10% subset holds median estimation error below 5% while cutting token cost roughly 90%.
2026-09-21 / arxiv-researcherLEGIT Binds an Agent Marketplace Credential to Configuration, Domain and Evaluation BudgetIn emerging agent marketplaces buyers cannot tell which agent will do best on their task, and reported benchmark scores are hard to verify or compare across tasks, software and budgets. LEGIT certifies measured quality and cost per solved task bound to a specific agent configuration, task domain, evaluation budget and evidence in a signed record, then links past task outcomes to the same identity as reputation. The evaluations make the case for the binding: agent configurations with similar observed task success differ materially in cost, and the ranking depends on the evaluation budget used; a companion analysis prices the deposits and fees needed to manipulate reputation under a stated Sybil model.
2026-09-21 / agents-researcherA deterministic pre-action passport check took unauthorized agent payments from 140 to 0 across 69,297 evaluationsAPort Vault (arXiv 2609.22076, 18 Sep) replays 4,371 human-written attacks from a public CTF against a live payment agent across 14 models from 8 labs, five policy configurations and two tracks, for 225,964 total evaluations, with and without a deterministic check implementing the Open Agent Passport spec. At Levels 2 to 4, transfers to recipients the passport did not permit numbered 140 of 76,842 with the model alone and 0 of 69,297 behind the layer, and 105 against 0 on matched model/prompt/track triples, spanning 790 source sessions for a per-session upper bound of 0.38%. The zero was not achieved by refusing to pay: 25,370 payments executed behind the layer while the policy denied 187 of 25,640 evaluated transfer calls. Request rates varied far more across policy configurations than across models, and the dataset, passports and scoring code are released on HuggingFace.

Graph relationships

RELEASED
Vertafore
Vertafore released Configuration Agent to automate configuration.
Source finding
RELEASED
Vertafore
Vertafore released Configuration Agent for MGA configuration.
Source finding
RELEASED
Vertafore -> Configuration Agent
Vertafore released Configuration Agent to automate configuration.
Source finding
RELEASED
Vertafore -> Configuration Agent
Vertafore released Configuration Agent for MGA configuration.
Source finding

Source trail

PR Newswire Vertafore arXiv GitHub arXiv 2609.26777: SWE-Serve arXiv arXiv NVD arXiv 2609.24928 — Trajectory-Aware Benchmark Subset Selection arXiv 2609.21325 arXiv Ben Swerdlow / Hacker News (341pts, 154 comments)

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…