---
title: "Grade a repo before an agent touches it"
description: "Ara scores every repository for autonomous work, then closes the gaps. Scope permissions, seed memory, and prove the loop builds — so the first run lands a pull request, not a mess."
url: "https://ara.so/product/agent-readiness"
section: "Platform"
---

# Grade a repo before an agent touches it

Ara scores every repository for autonomous work, then closes the gaps. Scope permissions, seed memory, and prove the loop builds — so the first run lands a pull request, not a mess.

## Three signals decide if an agent can ship here

### One score per repo

Build, tests, memory, and access graded together, so you fix the weakest link first.

### Least privilege

Pin the branches, paths, secrets, and tools an agent may touch before it runs.

### Prove the loop

Ara boots an isolated workspace and runs build, lint, and tests before the score clears.

## From cold repo to agent-ready in four steps

### Scan and score

Ara clones the repo, reads the build and ci config, and grades reproducibility, test signal, and access against the readiness rubric.

_Readiness scan_

### Scope access

Set the branches, paths, secrets, and tools the agent can use. Defaults are read-mostly; you widen them deliberately.

_Permissions_

### Seed memory

Ara drafts a repo wiki — architecture, conventions, gotchas — so the first run starts with context instead of guessing.

_Repo memory_

### Verify the loop

A sandbox run proves build and tests pass end to end. The repo flips to ready and clears for autonomous pull requests.

_Pull request_

## Score your repos before the first agent run

Point Ara at a repository and get a readiness grade, a scoped access plan, and a seeded memory draft in minutes.