Multi-Agent

PAIChecker finds PR–issue mismatch in 13.6% of SWE-bench Verified instances. Benchmark maintainers should validate that the issue, patch, and tests describe the same task before trusting agent scores.

Socher’s automated-research design combines prior knowledge, measurement data, simulation, physical experiments, and agent orchestration, with early demonstrations in training and CUDA optimization.

Vercel Sandbox can now isolate agents as Linux users while exposing selected files through a shared group directory. This gives multi-agent harnesses a simpler permission boundary.

Add agents beside deterministic fraud systems, not in place of them: project cross-service events into a semantic layer, send only uncertain cases to specialist agents, then emit the verdict back into the saga.

Controlled experiments suggest long-horizon agent planning depends on explicit state transitions, some compositional trajectories, and compatible teacher patterns—not atomic skills alone.

Fixed-rubric judges miss failures that emerge across long, variable agent trajectories. Arize argues for adding agent-based analysis while retaining deterministic and LLM-judge evals.

A fixed chain of specialist agents lost context and produced incoherent recommendations. The replacement separates deterministic detection, gives one agent end-to-end ownership, and uses subagents only for bounded investigations.

Cursor’s SQLite experiment suggests swarm harness design can matter more than model mix: separate planner and worker contexts reduced coordination churn while cheaper workers handled most tokens.

FormalAnalyticGeo shows a reusable synthetic-data pipeline: agents generate problems, compile them into a formal representation, render exact diagrams, measure answers, and retry failed checks.

A trending Claude Code framework (8.4k stars) that runs a job hunt end to end: /scrape ranks postings, /apply tailors LaTeX CVs, and a second reviewer agent plus a PDF-compile loop verifies the output.

Alberta's government ran 50 parallel Claude Code agents over 466M lines of code, compressing a security review estimated at 6.5 years into 20 hours — with every patch still gated on human review.

Dual-channel debates show social structure alone makes LLM agents diverge: public statements split from private ones, with decision divergence jumping from ~3% to ~40% across 10 models — no deceptive prompt needed.

Agora routes reasoning steps through an auction among expert models and tools, adding a single control for cost versus quality and outperforming matched baselines on five benchmarks.

Google's AMIE matched 21 primary-care physicians on longitudinal disease management in a blinded Nature study, scoring higher on plan preciseness and guideline alignment. Research-stage, not deployed.

A 100-person factory built a multi-agent sales system around organized company memory, specialist roles, and human approval. The useful pattern is retrieval and governance, not custom model training.

This Apache-2.0 collection provides runnable agent, skill, MCP, memory, multi-agent, and RAG examples across major model providers, useful for borrowing patterns before choosing a stack.

Osmani argues that agent throughput makes evidence-backed judgment the engineering bottleneck: builders should require explainable diffs, tests, logs, and explicit ownership before shipping.

Project Nanda proposes open discovery, identity and coordination layers for agents across vendors. Its local simulator lets builders test one protocol layer without building the entire network.

Model “done” as evidence, verification, authority, residual risk and next ownership—not an agent-set Boolean. This separates forward progress from approval in multi-agent workflows.

Claude Science (beta, June 30) packages 60+ domain skills, a coordinator/specialist/reviewer agent stack, and HPC/Modal compute into a research workbench with reproducible, auditable outputs.

OpenAI's plugin lets you drive Codex from inside Claude Code—slash commands for code review, adversarial critique, and delegating or handing off tasks to Codex background jobs. ~629 stars today.

Sixteen parallel Opus 4.6 agents wrote a 100k-line Rust C compiler in two weeks (~$20k) that builds Linux 6.9 — the writeup credits test quality and context hygiene, not raw model capability.

An Anthropic harness for multi-hour app builds pairs a generator agent with a Playwright-driven evaluator to counter self-grading bias — a $200, 6-hour run versus $9 solo, and it got simpler on Opus 4.6.

Terminal multiplexer for coding agents: a real terminal per agent across Claude Code, Codex, Copilot and 15+ others, with blocked/working/done state in a sidebar, persistent sessions, and a socket API.