Opus 5.5 agent flub: doesn't realize subagent transcripts are on disk
kalomaze · x · 2026-11-15
kalomaze shares a microcosm of his experience using Opus 5.5 as an agent: when asked about subagent progress, the model claimed it couldn't see partial findings since subagents only report at completion — then corrected itself, admitting their transcripts were on disk all along.
More from coding & agent
- Same prompt, two models: Opus 5.5 vs Fable 5.2 build Blender drone shots from scratch — airusk · 2026-11-15
- autoresearch@home: hundreds of AI agents collaborate on ML research via shared memory — GregoryDiamos · 2026-11-15
- Enterprise AI Needs Tools, Context and Evals, Not Omniscient Models — hwbhatti · 2026-11-15
- Prompt, context, harness, loop engineering are all just "knowing what you want" — mr_samosaman · 2026-11-15
- llm-memory: one SQLite file as shared memory for all your LLMs, via MCP — Mordicus1973 · 2026-11-15
- MCP server gives coding agents a designer: components rebuilt to match your brand — LuckyTNTyt · 2026-11-15