25launches

The first launch of AgentRun.

AgentRun page

AgentRun

Launching today

Open-source managed agents that learn from every run

Hunted by IlyasHunter
Visit agentrun.ai (opens in a new tab)

AgentRun is an open-source platform for running agents on Claude Code, Codex, OpenCode or pi, built to make repeat jobs cheaper. After each run, an agent can keep notes, a skill or a typed workflow that does the job without the agent loop; a grader checks them before they are committed to git. Runs survive server restarts, and an agent can get a Docker sandbox, a Linux desktop, a browser or an Android phone. Approvals, spend caps, one console. Apache-2.0; a hosted cloud is also offered.

Free optionsArtificial IntelligenceDeveloper ToolsOpen Source
1

Meet the team

Hunted by Ilyas. The makers are not on 25launches yet.

H (Hunter)
  • AgentRun screenshot 1
  • AgentRun screenshot 2
  • AgentRun screenshot 3
  • AgentRun screenshot 4

Discussion 1

Ask questions, give feedback. Comments don't count in the ranking.

Sign in to join the discussion.

  • IlyasHunter Pinned

    Agents are good at figuring things out, then figure out the same thing again on the next run, and you pay for every token. AgentRun lets an agent keep what it learned, as notes or a plain workflow, so a job it has done before costs less the next time. The team says its own code reviewer now costs 18 cents a review this way. It comes from Grep (YC F26), and the Show HN reached 91 points.