Meta launches personal agent Muse with sandboxed harness and Sentinel guard against prompt injection
shuchaobi · x · 2026-09-09
- Meta officially launched Muse, its personal agent (dogfooded internally since early 2026): it runs in the background, launches swarms of subagents, builds its own tools and edits itself, with access to your inbox, calendar and shell.
- Safety drove most of the engineering: the system assumes it may be under attack — the harness runs in an isolated cell, never sees real credentials, and every external interaction passes through a Sentinel the agent cannot override.
- The model was trained specifically for zero-shot tool calling via CLIs and skills, long context, long-trajectory instruction following with inherent prompt-injection awareness, and multi-agent coordination.
- Hardening came from extensive dogfooding, agentic red teaming, and a private bug bounty program; a detailed technical post accompanies the launch.
More from coding & agent
- LangChain details subagent forking as context engineering trick in deepagents — LangChain · 2026-09-09
- Box launches Mount to sync Box folders into agent sandboxes with built-in governance — badphilosopher · 2026-09-09
- Anthropic interviews WisprFlow, Actively and Pendo on building with Claude Managed Agents — ClaudeDevs · 2026-09-09
- Using ChatGPT Sites as a progress log for long-running agent projects — jdjohnson · 2026-09-09
- Muse review: agentic AI for normies with UI better than Anthropic or OpenAI apps — neil_chilson · 2026-09-09
- mumo MCP server routes questions across Claude, GPT, Gemini, Grok for cross-model debate — modelcontextprotocol · 2026-09-09