Opening an interactive Codex session on 0.157.0 now spins up a background app-server that survives after the TUI exits, copies ~374 MB of platform bits under ~/.codex, and runs an hourly updater that can force-kill active turns after a 60s drain. Opt out with codex features disable daemon_auto_start, then codex app-server daemon stop (and kill any leftover pid-update-loop).
OpenAI Codex lead Tibo raised GPT 6.1 Sol’s default context in Codex and ChatGPT Work from 272K to 500K tokens ( 80% more) to cut compactions on long tasks, with no long context fees on ChatGPT subscription usage. Override via model context window and model auto compact token limit in /.codex/config.toml ; the API max remains 1M…
TUI, web, and desktop now talk to a single local background service, every MCP server is wrapped as script functions in Code Mode so the agent can chain tool calls in one round trip, and subagents and shell commands can run in the background. Upgrade with opencode upgrade major from the latest 1.x; 1.0 plugins and the old server API no longer work…
Madhusudhan Vasanth Kumar’s Apache 2.0 terminal, VS Code, and Zed agent runs commands in Seatbelt (macOS) or bubblewrap (Linux) so they write only in the project, temp folders, and package caches, with home secrets unreadable and no network by default; edits show a diff first, and a hash chained audit log records every permission decision. Install with npm install g @garuda agent/garuda ; it works with Claude, OpenRouter, and local OpenAI compatible models…
Penn State, Montpellier, and Dartmouth showed LangGraph, AutoGen, Google ADK 2.0, AWS Strands, Microsoft Agent Framework, and CrewAI will execute a structured tool call if the model is prompted to echo it—even when the model later refuses. Their ToFu fuzzer also confirmed 253 bugs across 118 open source MCP servers, including 64 path traversal and three command injection flaws…
Xiamen University et al. open ProgramGym: 1,648 verifiable C/C++/Go/Rust environments built from GitHub repos for agents that rebuild programs from black box behavior, plus a role isolated Program Agent harness. Fine tuning Qwen3.8 27B on its trajectories raised ProgramBench from 34.87% to 42.84%; with Program Agent the SFT model hit 56.25%…
Spotted something we missed? Start a thread.