Anthropic's multi-agent system finds 66/70 planted bugs vs 14 for a single agent
Arindam_1729 · x · 2026-10-26
Anthropic launched Claude Managed Agents dynamic workflows in public beta. On a 116K-line codebase with 70 planted bugs, the multi-agent system found 66 bugs in all 3 runs, while a single agent found only 14, 15, and 27.
How it works: a lead agent creates a plan, splits work across multiple agents, runs them in phases, and combines findings — orchestrating up to 1,000 agents per run.
Try it: configure with multiagent20261001 and run via the Claude API managed agents endpoint.
More from coding & agent
- Even at max Daybreak tier, AI vuln-analysis requests still blocked by cybersecurity guardrails — jd_pressman · 2026-10-26
- Bambi quietly gains nonlinear model support, with a blog post showing how — twiecki · 2026-10-26
- Dev used Claude Opus 5.5 agents for 5 days to compose and render a 3D pianist cinematic — 141_1337 · 2026-10-26
- skillEd, an Apache-2.0 skill evaluation toolkit, asks users to test report reproducibility — MiserableFlatworm337 · 2026-10-26
- Fudan and Tencent unveil DraMaze, a multi-agent pipeline for diverse microdrama scripts — Siyang He · 2026-10-26
- Alibaba's MultiAgent-OPD trains hierarchical multi-agent systems end-to-end with on-policy distillation — TaoLiveAIGC · 2026-10-26