Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-09-09 · high
Resumed subagents and agent teammates paid full uncached input cost on every turn, with nothing failing to signal it.
Why now: Anthropic published the fixes in the 2.1.265 release on September 9.
Anthropic patched three prompt-cache bugs in Claude Code with the 2.1.265 release. None of them threw an error while they were live. The release notes for the fix call out cache-reuse breakage directly for two of the three.
First, resuming a subagent that started in the foreground changed its tool list and system prompt prefix. That mismatch invalidates the cache the moment the session picks back up.
The second hit agent teammates and resumed subagents specifically. It moved SubagentStart hook context and preloaded skills out of the prompt prefix on later turns. That knocked the cached tokens out of alignment, so every following turn reprices from scratch.
A third fix stops a resume from rewriting the last prompt after the process died mid-tool call.
For anyone running multiple agents or resuming long sessions, that meant turns after a resume were billed at full uncached input rates. Nothing failed. The session ran, the output looked right, and the only trace was a cost line higher than it should have been.
Each link below shares sources, entities, or timing with this story.
A Chinese lab shipped a runtime that manages two American coding agents as subagents, and it went from repo creation to 145,439 stars in four days. deepseek-ai/deepseek-harness published dsh-v0.1.0-rc.7 at 12:01 UTC today, its first tagged release since the repo appeared on Au...
Rust, created May 14, at 2,643 stars (GitHub). Every run produces checkpoints linking a commit to the session that made it, including prompts, tool calls and reasoning. It runs Claude Code, Codex, its own agent and anything from the ACP registry side by side against one codeba...
A GitHub repo cataloging Claude Code tips doesn't normally warrant a top story. But shanraisshan/claude-code-best-practice at 53.4K stars isn't a tips list anymore. It's the de facto reference for how an entire generation of developers is learning to work with AI coding agents...
Warp released its client codebase under AGPL-3.0, surged to 56,000 GitHub stars and #2 on GitHub Trending. But the real story isn't the open-sourcing. It's the repositioning. Warp isn't calling itself a terminal anymore. It's an "agentic development environment." The product n...
Two things in one release, and only one of them is in the security section. CLAUDE_CODE_SUBAGENT_MODEL used to override the model for every subagent. In 2.1.251 it sets the default instead. An agent definition's model: frontmatter now wins, and so does an explicit per-spawn mo...
1. Use claude agents --json to build session dashboards. Claude Code v2.1.145 outputs all live agent sessions as structured JSON with status, model, elapsed time, and parent relationships. Pipe it into a tmux status bar widget or session picker script for switching between bac...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.