Now on npm latest, the release flips the command sandbox to allow local network access, so agent-run commands can hit local databases and admin pages unless you lock it down, and adds --trust-folder for one-off -p runs plus a per-session /sandbox toggle. It also stops sessions stalling when an MCP server hangs while loading tools (AI Dev Pulse).
Released Oct 20, it now retries without tool search when a provider or gateway rejects it, and adds sandbox.allowNestedUserNamespaces so Linux sandboxed commands like Chromium or rootless podman can run, at the cost of bypassing the sandbox's file protections. It also fixes bare repo worktrees starting from local HEAD instead of the default branch and stops workflow resumes re running finished pipeline() and parallel() agents…
Since Oct 20, GPT 5.6 and later models in the Responses API accept prompt cache options.ttl: "12h" (or a per breakpoint ttl ), with 12 hour cache writes billed at 2.5x input versus 1.25x for the default 30 minutes and reads unchanged (changelog). OpenAI's Tibo (@thsottiaux) added that caches for logged in Codex and Work users will be kept permanently, so resuming an old conversation hits the cache (PANews)…
Released Oct 20 with FP8, NVFP4, GGUF, and MLX variants, the reasoning model has a 262K context and serves via vLLM or SGLang with tool calling, and Ornith reports 0.845 on SWE bench Verified and 0.813 on Terminal Bench 2.1 (its own numbers, per HuggingNews). Community quants from AtomicChat bring it down to about 35GB for 64GB Macs…
Per The Information, a tester said the unreleased model spotted broader related problems when asked for a simple bug fix, but it's unclear whether it can tell an incomplete request from work you deliberately left out. That gap matters: the Overeager Coding Agents study found Claude Code, Codex CLI, and Gemini CLI took out of scope actions in 5.4% to 27.7% of runs, versus 0.2% to 4.5% for ask first OpenHands…
Calling /create formal verification skill makes the agent pick a high stakes property (races, retries, parsers, unsafe code), choose a tool such as TLA+, CBMC, or Kani, show the check failing on a deliberately broken variant, then leave a project local skill so the next agent can rerun it with one command. pstack also ships into Claude Code's plugin marketplace from the same repo, and Cursor's own eval behind the skill is small (one run per repo on six open source projects)…
Spotted something we missed? Start a thread.
