Tag
9 posts tagged.
An MCP stdio server that holds a timer or a pool open does not exit when its client dies. Measured on SDK 1.30.0 and Node 24.8.0: it is reparented to init and keeps running. StdioServerTransport registers no stdin end-of-file listener. Here is the reproduction and the four-line fix.
You gave your agent a shell tool, used execFile so there is no shell, and allowlisted the programs. It is still not safe. If the agent controls the argument array, tar's own --checkpoint-action=exec runs any command with no metacharacters. Here is the reproduction and the per-binary argument policy that stops it, measured on Node 24.8.0.
If your agent has a read_file tool, you probably guard it with path.resolve plus startsWith. It blocks the obvious traversal and is defeated by a symlink and by a sibling directory sharing the prefix. Both reproduced on Node 24.8.0, with the guard that passes all nine cases.
Set _meta["anthropic/requiresUserInteraction"] to true and Claude Code prompts for that tool on every call, even in bypassPermissions, and even when an allow rule matches. It shipped in v2.1.199 with no changelog entry, which is why almost nobody has written about it.
Four launch-week tutorials carried a stored publication timestamp that predated the registration of agentnotebook.dev. The timestamps are corrected, the tutorial text is unchanged, and the check is one you can run yourself.
A hands-on 2026 tutorial: stand up your own self-hosted, open-source AI app builder from the ai-app-builder-open repo, with one API key and code you fully own.
A runnable comparison of the Claude Agent SDK and LangGraph in 2026, with the same agent built in both and an honest decision matrix for when each wins.
Make a Claude turn survive a browser reload: server-side ring buffer keyed by message.id, the WHATWG Last-Event-ID header, and a Next.js App Router Route Handler that replays missed SSE frames in 180 ms. Measured June 2026 on claude-sonnet-4-6.
