Brain · Loop Library
Loop Library
The Loop Library is a browsable, copy-paste gallery of named agent loops. Each card gives you a paste-ready command, a spend and iteration cap, a verifier requirement (a separate checker, never the maker grading itself), and an estimate of what that loop costs on your own telemetry. You copy a loop into your own tool. RepoOps hands you the loop and its cost and runs nothing.
Where to find it
- Local tab:
/loop-library.html?repo=<id> - In the app:
Brain > Loop Library - Gallery API:
GET /api/loop-library?repo=<id> - Save API:
POST /api/loop-library/save?repo=<id>&loopId=<id>
What it does for you
Find a loop by task or categoryA search box matches on name, task, category, and verb, and category chips (Engineering, Evaluation, Operations, Content, Design) narrow the grid. The gallery is a curated seed set, so you scan for the loop that fits the work in front of you instead of writing one from scratch.
Copy a paste-ready command into your own toolEvery card carries a command you can copy with one click and paste into your own agent. This is browse and copy only: RepoOps emits the loop and does not execute anything on your behalf.
See the cap, the verifier, and what it costs hereEach card shows a budget cap (spend plus iteration ceiling), the verifier that checks the loop's work, and a 'Costs here' estimate scaled from your own session telemetry. When nothing is attributable locally the estimate reads 'unpriced', never a fabricated $0.
Save a loop into your repoSave to repo writes the loop as .claude/brain/loops/<id>.md in the selected tracked repo, keeping the name, category, cap, verifier, and command as a markdown file you can version and edit.