Claude Code 2.1.261 full changelog: 128K output cap, skill-doctor, interrupt fixes
ClaudeCodeLog · x · 2026-09-05
Full changelog for Claude Code CLI 2.1.261 (67 changes):
New features
- bashOutputMaxChars and taskOutputMaxChars settings raise inline command/background-task output up to 128K characters before saving to file.
- /skill-doctor shows which loaded skills go unused and what they cost in context, enabling pruning.
- --append-subagent-system-prompt-file reads oversized subagent system prompts from a file.
- /status and claude doctor now explain why organization policy couldn't load (e.g. proxy not passing the endpoint).
Fixes
- Typed/pasted characters landing out of order or dropped during fast input or key repeat.
- False "couldn't be resolved" error from /add-dir <subdirectory> and more.
More from coding & agent
- unlazy: open-source skill fights AI agent laziness with runnable gates — tom_doerr · 2026-09-05
- Codex's new voice mode picks up sniffing, tongue clicks, coughs and gulps in real time — altryne · 2026-09-05
- A 7-Step Roadmap for Engineering Production-Ready AI Agent Personas — MaryamMiradi · 2026-09-05
- Reef launches inference-native infra that serves self-improving agents without downtime — pliang279 · 2026-09-05
- OpenAI Employee Uses GPT-6 Astra in Codex to Edit Official ChatGPT Instagram Videos — craigsdennis · 2026-09-05
- Gemini 3.8 Flash beats larger models on agent benchmarks, built for cheap scale — VraserX · 2026-09-05