AI Discovery workflows

Evidence-gated compositions for launch review, generative visibility, and agent action readiness.

Compositions are consultative routes, not an execution engine. Skip any step whose output is not needed.

Pre-launch review

1. Represent: stabilize people, brands, methods, offerings, claims, and authoritative sources.
2. Inspect search foundations: evaluate intended access, indexability, canonical signals, architecture, and structured representation.
3. Design measurement: define the sample, raw records, metadata, variance, and comparison rule.
4. Review generative visibility: only after repeated observations exist, compare source presence, citations, extraction, and fidelity.
5. Handoff: assign each proposed change and recheck to an owner.

Stop when an entity decision, access permission, publication approval, or comparable baseline is missing.

Generative visibility review

Use ai-discovery-measurement before generative-visibility-optimization. Preserve prompts, outputs, citations, run context, and unavailable fields. A missing citation is an observation, not proof that a source cannot be retrieved.

Agent action readiness

Use agent-capability-actionability to define the user task and control contract. Pair it with ai-discovery-measurement to observe task completion, errors, confirmation, side effects, and recovery. Do not introduce MCP, A2A, WebMCP, or OpenAPI unless the task and existing interfaces justify them.

Useful dependencies only

The capability overlay represents four relationships: entity representation informs search; search informs generative visibility; measurement informs generative visibility; actionability informs measurement. These links do not activate skills automatically.