Trello

Authentication traditionally uses API key plus token flows, with Atlassian ownership adding some ecosystem nuance. The model is simple enough for agents, but not especially elegant by modern OAuth-first standards. Access is primarily shaped by board/workspace permissions, so the real safety boundary is what the token-bearing identity can access.

The docs are long-lived, practical, and sufficient for real integration work. Trello's simplicity helps a lot here: agents can typically become productive with the API quickly, and there are few hidden concepts compared with heavier project-management tools.

Trello remains a strong agent surface because its model is simple and human-readable: boards contain lists, lists contain cards, and cards carry members, comments, labels, due dates, and attachments. Agents can use Trello as a lightweight workflow system where humans still need to inspect or move work visually. It is not as deep as Jira or Linear, but that simplicity is often a feature.

The API is mature and straightforward. Boards, lists, cards, custom fields, members, labels, actions, and attachments are all exposed in predictable REST shapes. This makes it easy for agents to create cards from events, move cards through workflow stages, sync metadata, or extract board state for reporting. The data model is intentionally uncomplicated, which reduces automation overhead.

Reliability is generally high, but the main operational issues are workflow ambiguity and human-agent collisions: users may move cards manually, rename lists, or repurpose boards in ways that break automation assumptions. Trello itself is stable; the fragile part is usually the team process encoded in board structure rather than the API mechanics.