Declare · Declared intent
Declared intent
What a loop is declared to do, before it runs. Reads each declared loop's contract - its goal, the check that proves it done, and the stop rules - and joins the loop registry's run history so a declared loop links straight to its outcomes: the prompt-to-outcome spine, read left to right.
Where to find it
- Desktop:
http://localhost:4000/, then Declare → Declared intent in the sidebar. - Hosted:
repoops.ai/team/declare - Keyboard: ⌘ K, then type “Declared intent”.
- API:
GET /api/declare?repo=<id> - Source:
.claude/loops/<id>.loop.md(the loop contract)
What it does for you
Every loop's contract, in one place.Goal, the check (command + pass condition) that proves it done, and the stop rules - read straight from the loop contract file, never re-typed.
Honest empty states, not blank pages.A loop with no contract reads uncontracted; a contract with no runs reads contract-only. A repo that declares neither loops nor contracts says so plainly.
RepoOps reads and judges; it never runs the loop.The contract is the source of truth for what SHOULD happen. RepoOps never executes it - it only reads the contract and joins it to what actually happened.
Configure
Nothing to configure. Author a contract with the loop-spec skill, or hand-write .claude/loops/<id>.loop.md. LLM-free.