# Advisory AI agent guide

## Mission
Advisory AI is the retrieval-augmented assistant that synthesizes advisory and VEX evidence into operator-ready summaries, conflict explanations, and remediation plans with strict provenance.

## Key docs
- [Module README](./README.md)
- [Architecture](./architecture.md)
- [Implementation plan](./implementation_plan.md)
- [Task board](./TASKS.md)

## How to get started
1. Review ./architecture.md for retrieval pipeline, guardrails, and profile support.
2. Open ../../implplan/SPRINTS.md and locate stories for this component.
3. Check ./TASKS.md and update status before/after work.
4. Read README/architecture for design context and update as the implementation evolves.

## Guardrails
- Uphold Aggregation-Only Contract boundaries when consuming ingestion data.
- Preserve determinism and provenance in all derived outputs.
- Document offline/air-gap pathways for any new feature.
- Update telemetry/observability assets alongside feature work.

## Required Reading
- `docs/modules/advisory-ai/README.md`
- `docs/modules/advisory-ai/architecture.md`
- `docs/modules/advisory-ai/implementation_plan.md`
- `docs/modules/platform/architecture-overview.md`

## Working Agreement
- 1. Update task status to `DOING`/`DONE` in both `docs/implplan/SPRINTS.md` and the local `TASKS.md` when you start or finish work.
- 2. Review this charter and the Required Reading documents before coding; confirm prerequisites are met.
- 3. Keep changes deterministic (stable ordering, timestamps, hashes) and align with offline/air-gap expectations.
- 4. Coordinate doc updates, tests, and cross-guild communication whenever contracts or workflows change.
- 5. Revert to `TODO` if you pause the task without shipping changes; leave notes in commit/PR descriptions for context.