Pagination

Runs, plans and reviews are listed newest first, one page at a time. Each page carries a cursor for the next one:

1. Call the list endpoint, optionally with limit.
2. If next in the answer is not null, call it again with cursor set to that value.
3. Stop when next is null.

GET /api/runs?limit=20

{
  "runs": [ { "run_id": "01JB8X6F4M2C9QZ3T7V5W1K0HA", "status": "completed" } ],
  "next": "01JB8W2Q7N4D6R8S0T2V4X6Z8A",
  "scanned": 20
}

GET /api/runs?limit=20&cursor=01JB8W2Q7N4D6R8S0T2V4X6Z8A

Cursors are opaque: pass them back as you got them, and don't build your own. They don't expire.

GET /api/repos is not paged: a team's repositories come back in one answer.

Short pages when searching runsCopy Anchor Link

GET /api/runs with a search (q) or filters (status, source, repo) reads a few hundred runs per request and keeps the ones that match. A page can then hold fewer than limit runs, or none, while next is still set: older runs are unread. Keep reading until next is null. scanned says how many runs the page read.

Reading every pageCopy Anchor Link

async function* allRuns(query: Record<string, string> = {}) {
  let cursor: string | null = null;
  do {
    const params = new URLSearchParams({ ...query, limit: '100', ...(cursor ? { cursor } : {}) });
    const res = await fetch(`https://api.coderhelm.com/api/runs?${params}`, {
      headers: { Authorization: `Bearer ${process.env.CODERHELM_API_KEY}` },
    });
    if (!res.ok) throw new Error((await res.json()).error);
    const page = await res.json();
    yield* page.runs;
    cursor = page.next;
  } while (cursor);
}

for await (const run of allRuns({ status: 'failed' })) console.log(run.run_id, run.title);

GET /api/plans also returns its cursor as next_cursor, kept for older clients; use next.