What a health check looks like when you're working with agents.
![Screenshot of the SAM (Simple Agent Manager) chat interface. A green system message labeled "[SAM Orchestrator Check-In]" warns that the task appears to have stalled due to several minutes of inactivity. It instructs the agent to send a brief progress update and continue working if unfinished, call complete_task() if finished, or request_human_input() if assistance is needed, warning that otherwise the task will be marked as failed.
Below it, the assistant replies: "The work is finished: PR #1436 was pushed, checked, and merged successfully. I'm sending the final SAM task completion signal now so the orchestrator does not mark it stalled."
At the bottom is a collapsed tool invocation labeled "Tool: sam-mcp/complete_task" with a green success indicator, showing that the completion signal was sent. The left sidebar displays a list of previous chat sessions.](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:lystr3dpdy3daz2arrtrztbi/bafkreicndmc6wmaheiiscgx2mbibdvu7qo2iw2aorclwpayyq66wlaptkq)
Simple Agent Manager
@simple-agent-manager.org
An open source platform to run many agents at a time. Control plane built on CloudFlare, with your agents running on VMs in your cloud provider of choice. Clouds: Hetzner, Scaleway, GCP Agents: Claude Code, Codex, Vibe, OpenCode, AMP Built in Paris 🇫🇷
Feeling loopy. And I'm not even sure what that meeeans.
Made it easier to setup your own deployment. Light mode looking half-decent too. #ai #opensource #claude #codex #cloudflare
Uber reportedly now caps coding agents at $1,500/month per employee per tool - seems sensible to me, but it's also an interesting hint at the value Uber thinks these tools are providing simonwillison.net/2026/Jun/3/u...
Uber Caps Usage of AI Tools Like Claude Code to Manage Costs
I wrote the other day about Uber blowing its 2026 AI budget in four months, and how that wasn't particularly surprising given they would have set that budget in 2025, …
simonwillison.net
