Knowledge search

Data

Builds reusable retrieval agents that answer from a curated document corpus — embeddings, a search index, a knowledge library, schema and header mapping, reuse of prior decisions, confidence thresholds and a human review loop.

Used when

When a question should be answered from a defined corpus rather than from a model's general knowledge, and the answer needs a confidence and a reviewer.

Loads first

• The retrieval-system and agent-building skills.
• The shared engineering guidelines.

Returns

• Whether the use case is retrieval at all, and why.
• The reusable agent boundary and its runtime contract.
• The corpus and metadata contract.
• The retrieval and confidence policy.
• The production stack and the local emulation path that replays it.
• The human review and learning loop for runtime decisions.
• The evaluation and operations plan, and an implementation checklist.

Hands off

• Search to production — taking a retrieval system from prototype to production