Fetching from the wire…
Tools2026-09-13 · source-backed
PR #29214, merged September 11 and in nightly v0.61.0-nightly.20260912, stops --sandbox runs mounting the host user's persistent configuration and credentials. It replaces host directory mounts with sanitized config files, standardizes on realpath resolution during path-sensitivity checks with a fallback for non-existent paths, validates resolved working directories against sensitive host locations before mounting, and restricts ephemeral directory permissions synchronously. Covers Docker, Podman, runsc, LXC and macOS sandbox-exec Seatbelt profiles. If you were running --sandbox and assuming your credentials weren't in there, they were.
Each link below shares sources, entities, or timing with this story.
v1.9.0, released September 8, adds --allow-unrestricted-paths, configurable filesystem roots, an option to disable the JS execution tools, and extends --no-javascript-evaluation to cover navigations and initScripts. It also raises the default trace buffer to 1.2GB to match Dev...
tokentab (September 7, 481 stars) computes cost from Claude Code, Codex and Gemini CLI logs. tracecrate (September 10, 105 stars) is a local-first workbench inspecting Claude Code, Codex and OTLP traces and comparing runs with no backend or keys. ai-data-extractor (September 1...
There's been a fundamental gap in AI coding workflows that's been driving me crazy: your agent writes React components, generates CSS, builds entire UIs, but can't see the result. It's coding blind. The Chrome DevTools MCP server from Google's ChromeDevTools team fixes this, a...
MIT-licensed desktop app and CLI that auto-detects installed AI clients and manages MCP server configuration for all of them — Claude Code, Claude Desktop, Cursor, VS Code, Windsurf, ChatGPT Desktop, Gemini CLI. Built-in MCP marketplace, team snapshot export, automatic backups...
5.8k stars, Rust, macOS, ~40 MB versus ~67 MB upstream, keeping full Lua customization. Command-failure recovery with suggested fixes applied via Cmd + Shift + E, natural-language-to-command via # <description>, preconfigured integration for Claude Code, Codex, Gemini CLI, and...
wanshuiyin/HERO-Anti-OverDefense went from creation to 68 stars in a single day. HERO is Hashing, Edge cases, Rubrics, Overbuild, and the claim is that agent over-engineering isn't diffuse but falls into four recognizable shapes suppressible with a portable prompt contract acr...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.