Anthropic's Fix: 2 Agents + 3 Files for Cross-Session Memory
sujingshen · x · 2026-07-03
Anthropic's solution to the Agent memory problem is surprisingly simple: an initialization Agent breaks down tasks (like cloning claude.ai) into a 200+ item feature list, each marked "failed". Every time the coding Agent starts, it reads three things—the feature list (what to do), a progress file (where it left off), and the git log (correctness). It then executes exactly one task, commits, updates progress, and passes the state to the next Agent. The entire flow relies on files, not memory, to transfer context.
More from coding & agent
- Agent verified 'success' after double-voting: a lesson in harness vs model bugs — TejasKumar_ · 2026-10-26
- diginomica: agentic automation hype is overdone, and 'The Hackening' of enterprise security has begun — jonerp · 2026-10-26
- 'Vibe verification': ~60% of AI-written formal specs prove the wrong property — basedjensen · 2026-10-26
- Devin launches Security suite powered by Claude Mythos with $100M+ credits — DevinAI · 2026-10-26
- HackBench injects reward hacks into agent trajectories to stress-test monitors — DanielKhashabi · 2026-10-26
- Hugging Face ships hf CLI updates: deduped cache saves 19GB, faster buckets, retryable GPU jobs — yb2698 · 2026-10-26