Runs Claude Code, Codex, OpenCode or pi agents with durable execution, spend caps and approvals, and makes them learn notes and skills from each run.
Use it now
Keep it with up to 9 other tools you're setting up, then share them all as one link.
Try this first
Ask your coding agent:
- Durable Runs
Use AgentRun to store every run: ‹what you need›
- Typed Workflows
Use AgentRun to run repeatable work: ‹what you need›
- Approvals and Spend Caps
Use AgentRun to pause runs: ‹what you need›
Install the CLI
Install it as its own instructions say.
Setup guide: agentrun.ai/docs/quickstart
Call the API
Auth An API key from Parcha
Setup guide: agentrun.ai/docs/quickstart
Compare
Press Compare to add it to the tray at the bottom of the window, then pick up to 2 others anywhere on the site.
- Access
- CLIAPI
- Auth
- API Key
- Skills
- 5
Skills
Learn From Each Run
Keeps notes, skills and learned workflows an agent proposes after a run, checks and grades them, and commits the keepers.
Durable Runs
Stores every run so it carries on after a server restart, inside a sandbox, Linux desktop, Android phone or browser.
Typed Workflows
Runs repeatable work as typed workflows of tool calls, model steps and scored decisions, falling back to an agent when unsure.
Approvals and Spend Caps
Pauses runs for human answers and approvals, caps each run's cost, and shows every agent's runs and spend in one console.
Agents as an API
Starts and waits on agent runs over HTTP or with the TypeScript and Python clients, using a server token for auth.