Continue
Every agent can see every chat in the project. Start in Codex, pick it up in Cursor, with nothing pasted across.
Use ACM to continue from Codex.
MCP server + CLI for coding agents
Tired of copying a chat into the next agent? Pick up the same task where you left off, or tell one agent to put the others to work.
01Interactive demo
Follow a simulated run with Claude Code, Codex, Cursor and OpenCode. Switch to Herdr to see the same work in its terminal layout, or click a pane to look closer.
You ask Claude. One prompt, three jobs.
~/acme-api · main
idle · waiting for a brief
idle · waiting for a brief
idle · waiting for a brief
src/middleware/rateLimit.tsfreesrc/routes/login.tsfreetests/login.rateLimit.test.tsfreedocs/api/auth.mdfree02What it does
Every agent can see every chat in the project. Start in Codex, pick it up in Cursor, with nothing pasted across.
Use ACM to continue from Codex.
One agent briefs the others. Each one claims the files it edits so they never collide, then reports back.
Tell codex to write plan.md, and claude to write packing.md.
Track a task, record a decision, and find it again later, whichever agent you happen to be in.
Use ACM to remember we chose Postgres over SQLite.
03Works with
--permission-mode autoSafe actions run on their own. A denied one comes back to you as needs_approval.
--approve-for-meApproves its own safe steps. Anything riskier waits for you.
--auto-reviewAuto-review runs safe actions and stops on the rest.
--autoNo auto-review mode, so it approves anything not explicitly denied.
These are the agents supported today and featured in the demo. More agent integrations are planned.
With herdr installed, every delegated agent gets its own pane you can watch: herdr session attach acm