Saltar al contenido principal

Atlas AI

Atlas AI supports research and implementation decisions using AtlasRepo's published catalog and knowledge projections. It can search solution records, save or archive account-scoped work and query supported knowledge endpoints. It does not deploy software, approve dependencies or replace review of upstream sources.

How to use it​

Give the assistant an outcome, constraints and the evidence you need. A useful request includes hosting preference, language or runtime, security constraints, expected scale and a stop condition. Review every proposed project against its source before acting.

Public OpenAPI operations currently cover solution search, saved solution work, dossiers and knowledge queries under /api/assistant/.... Authentication and request bodies differ by operation; generate integrations from the live public specification rather than guessing fields.

If catalog or knowledge data is unavailable, treat the result as incomplete. Do not silently substitute a stale or invented recommendation.

My solutions history​

When Atlas AI is enabled, /member?panel=solutions shows private conversations on the left, questions and answers in the centre, and the selected answer's sources on the right. Reopening a saved conversation does not generate another answer. The server checks current access and source versions before showing saved evidence; withdrawn or changed sources can make an older answer unavailable.

Questions and answers are stored in your account, not added to the public search corpus. Each new question is independent: previous messages are not automatically sent as model context. The submitted question and relevant source excerpts may be sent to the configured AI provider. Archive hides a conversation; it is not permanent deletion. Failed or uncertain submissions are not automatically regenerated.