Gemini CLI 0.61.0 (Sept 24) now requires confirmation before editing recognized build files (package.json, Makefile, pyproject.toml, Bazel BUILD), before later build/test commands after those edits, and before shell commands whose args match untrusted web/MCP content—with no “always allow.” The same release stops mounting ~/.gemini into the sandbox and strips API keys and hooks from the in-sandbox settings copy.
Announced Nov 10 as in development, it lets compatible AI clients run cross system investigations and request supported actions like access changes, even on legacy platforms without MCP. Defined policies decide what runs automatically and what needs human approval…
Announced Nov 11, it goes beyond search: sellers can draft follow ups, experts can save answers to projects, and teams can tag and route new content for review. Every action respects the user's existing Responsive permissions…
Announced Nov 11, it gives Claude a code knowledge graph of your repositories to find callers, trace symbols and check what a change reaches, with file and line citations. Claude Code can add it with claude mcp add transport http scope user graphify https://api.graphify.com/mcp…
Announced Nov 10, it lets an AI assistant build decks from connected sources like Tableau, Power BI and Snowflake, then export to PowerPoint, Google Slides or PDF. It can also refresh, schedule and email decks, with figures traced back to their source…
Announced Nov 10, sentry mcp starts the MCP server in stdio mode and shares the credentials from sentry auth login , so the config needs no token. Example: claude mcp add sentry sentry mcp ; the OAuth HTTP endpoint is still available…
Spotted something we missed? Start a thread.