AgentRun
Open-source DSL that freezes repeated agent work into typed, inspectable steps
Emerged finds new products worth a look, daily — what’s new today
AgentRun
A layer over Claude Code, Codex or OpenCode that keeps lessons from each run
If your agents repeat the same expensive task daily, this turns what worked into checked notes, skills and workflows, so later runs cost fewer tokens.
AgentRun sits on top of an existing agent harness such as Claude Code, Codex, OpenCode or pi. After each run the agent proposes what to keep; code checks and a grader filter it, and survivors are committed to the agent's git folder, with risky rules held for approval. Runs survive server restarts, and a console handles approvals, spend caps and monitoring. Start it with npx, Docker Compose or Render. From Parcha Labs (Grep.ai). Subscription logins are not supported yet.
More like this
OpenBot
Self-hosted AI coworkers, each with its own computer and a policy gateway
OtoDock
Self-hosted company OS that runs Claude Code and Codex agents as staff
Agent Autoloop
A local scheduler that keeps coding agents working and asks you only when stuck
DeepSeek Harness
DeepSeek's open-source agent harness where every capability is a plugin
Omnara
Durable AI agents you can self-host, on any model and any machine