Claude Code v2.1.274: self-healing transcripts and a batch of MCP fixes
ashwin-ant · ghdev · 2026-09-17
anthropics/claude-code shipped v2.1.274 with notable fixes and features:
- Stability: corrupted transcripts that caused endless "unexpected tooluseid" 400 retries now self-heal, or fail fast with a clear error and a /rewind hint
- MCP fixes: legacy HTTP+SSE servers failing on 4xx responses, Streamable HTTP tool calls hard-timing out at 5 minutes, prompts/resources not refreshing without a listChanged declaration, and 403 insufficientscope errors misreported as expired sign-in (now names missing permissions)
- New config: CLAUDECODEMCPSTARTUPWAITMS to bound first-turn MCP wait; configurable Postgres connect timeout (default 5s) in the Claude apps gateway
- Observability: effort attribute on OTel spans plus a managed-settings resolution event
- UX: visible critical-memory warning with recovery steps, and click-to-expand for collapsed teammate/agent messages
More from coding & agent
- LangChain's Jev targets constrained output for building agent harnesses — multiply_matrix · 2026-09-18
- NewEyes AI brings a hybrid on-device/cloud visual agent to Meta glasses — rohanpaul_ai · 2026-09-18
- Cursor Projects wins praise: kanban tracking, multi-model agents and 3-way bakeoffs — kieranklaassen · 2026-09-18
- OpenAI's Astra Struggles With Long-Term Architecture: Duplication, Overengineering, Bad Assumptions Linger — Substantial_Swan_144 · 2026-09-18
- Hang Ten Systems raises $53M more in seed funding, totaling $85M — vsikka · 2026-09-18
- Does AI coding actually become multiplayer? Shared Slack agents vs. supercharged solo devs — Telos_in_the_Void · 2026-09-18