[gormaniac]

get_profiles() and get_users() called _model(). That passes the response to convert_to_model(), which turns {"data": [...], "next": ...} into a bare list and drops next. Triage paginates every collection: the default page is 50 and the max is 200, per https://tria.ge/docs/cloud-api/conventions/. As a result:

• Callers silently got only the first page.
• The offset argument was unusable, because it needs the next value of the previous page, which callers never saw.

The slib.hatching Storm service (gormaniac/stormsvcs#14) currently works around this by calling the private _request() itself.

Changes

• Paged methods: get_profiles, get_users, and get_user_apikeys now follow next in pages of 200 and return every item.
  • limit caps the total; None or 0 returns all.
  • offset starts from an earlier page.
  • This reuses the _pages/_list/_iter helpers the sample lists already use.
• New methods:
  • profiles_page and users_page return a single page as a new base.ListPage, with data plus next.
  • iter_profiles and iter_users stream items across pages, matching iter_samples/iter_search.
• Errors: handled the same way as before. An ErrorResponse is returned, or raised with raise_on_api_err; the iter_* methods raise.
• FakeTriage: now pages like the docs describe (default 50, max 200) and paginates API key lists too.
• CHANGELOG: entries added under Unreleased.

Behavior change: calling get_profiles()/get_users() with no limit used to return the API's first page (50 items). It now returns everything. Code that relied on getting a single page can use profiles_page/users_page.

Testing

• pytest tests: 45 passed, 4 skipped (the live tests; no HATCHING_TOKEN was available).
• New tests cover:
  • 251 profiles across two max-size pages, plus limit.
  • Single pages and starting from an offset.
  • The iter_* methods.
  • 61 users and 210 API keys.
  • Auth errors, both returned and raised.
• Not run against the live API.

---

Generated by Claude Code