The TUI no longer updates itself: run /update to install a new version or move to OpenCode 2, or use opencode upgrade --major from the shell. The release also allows OpenAI's ultrafast service tier and checks PTY connection origins and tickets before anything else.
Every Veda app declares typed actions (70+ across the built in apps) that the agent runs through the same code path as your clicks, and the OS itself, not the model, holds sensitive or destructive actions until you approve them at the desktop. It's GPLv3, runs in QEMU, VirtualBox, or from a USB stick, and needs a Deepgram key for speech and the language model…
The Core AI team's canonical plugin exposes any public WordPress Abilities to agents at /wp json/mcp/mcp adapter default server and lets developers build custom servers via mcp adapter init . Version 0.7.0 adds the stateless 2026 07 28 MCP spec while older clients still connect, and the plugin remains experimental ahead of v1.0…
The free, open source Tauri app shows every agent's live status across your machines, turns pending commands into allow/deny cards, and runs scheduled routines, talking to agents directly with no account or telemetry. There's no binary yet: it builds from source with a one line installer on macOS 13+…
Name a project or describe what it does and Replit can pull its context into the current chat, so you can port a feature between apps or draft docs across several sites at once. Changes run as background tasks you review, and nothing is published automatically…
Auto review now asks for approval before recursive, force, or wildcard deletes whose target could resolve differently than written, like an unset variable in a path, and /skills warns about skills that break Agent Skills limits such as a mismatched name or a description over 1,024 characters. The release also adds Claude Opus 5.5 and Sonnet 5.5 on Bedrock and fixes ACP clients never getting wakeups from watched background shells…
Spotted something we missed? Start a thread.
