Announced Oct 22, the permission-aware server that launched in Claude in August now also answers intranet questions in ChatGPT, Microsoft Copilot Studio, and Slack, with each employee signing in as themselves. From Oct 26, assistants on all four can also draft private knowledge pages and draft or publish group posts, an optional capability admins can block per tool.
Released Oct 23, it saves reusable prompts and slash commands in ten numbered /presets slots, lets you set model, reasoning effort, and permission profiles inline with /model and /permissions , and authenticates enterprise MCP connections via Codex account grants or dedicated sign in. It also masks credentials in remote Bash and Zsh snapshot sessions and stops failed compaction from leaving partial summaries in history…
Released Oct 21, it fixes environment variables, including secrets, passed to external agents and MCP servers in remote projects showing up in the local machine's process list, and stops prompts to external agents stalling on needless Git checkpoints. It also adds context compaction and session notices for ACP agents, new Bedrock models, and binaries about 33% smaller on macOS…
Released Oct 22, it stops sandboxed Bash from running shell startup files outside the sandbox, blocks sandboxed commands and auto accepted edits from rewriting the mcp config file, and makes bypass mode rm prompts in Desktop and VS Code deny at once so unattended sessions keep going. claude ultrareview now asks for workspace trust on a terminal and refuses otherwise, so scripts need trust workspace…
Announced Oct 22, the upgraded server at api.causely.app/mcp ships the guidance agents need to use its tools on first connection and ties every agent action back to the diagnosis and evidence behind it, for Claude Code, Cursor, Codex, and other MCP clients. Causely's own benchmark claims 48% fewer tokens and 63% faster answers, and background agent examples are open source at causely oss…
Announced Oct 22, the Git host now runs an MCP server at https://mcp.codefloe.com/mcp that Claude Code, Codex, VS Code, or any MCP client can use to work with your repositories, issues, pull requests, releases, Pages, and CI quota. It authenticates with your own access token, so the agent can do only what that token's scopes allow…
Spotted something we missed? Start a thread.