The onion-intelligence repository hosts the core intelligence engine of the FLATLINEDSTAR ecosystem. It provides high‑level decision‑making, analytics, and policy logic that consumes observations from the data pipeline and produces actionable insights.
Planning – the repository currently contains only a placeholder README. No production code exists yet.
It separates intelligence (the reasoning layer) from the platform (deployment/runtime) allowing the engine to be versioned, tested, and used independently (e.g., by the onion-cli).
- Consumes normalized data from query‑engine (and indirectly from
tech-fingerprint,entity-extractor,change-detector). - Exposes a stable Go/Python/JS SDK via
onion-sdkfor downstream consumers. - Provides a set of public interfaces such as
OnionTarget,OnionObservation,SearchQuery,SearchResult.
- Define core data models and interfaces.
- Implement rule‑engine skeleton.
- Provide a JSON‑RPC API for query execution.
Not applicable until implementation.
- Language: Go (primary) with optional bindings for Python.
- Follow the contribution guidelines in the organization
.githubdirectory.
- Unit tests for each rule module.
- Integration tests using mocked observations from downstream services.
See the organization‑wide CONTRIBUTING.md.
- Phase 1: Define data contracts (shared types in
onion-sdk). - Phase 2: Implement rule engine core.
- Phase 3: Expose public API and integrate with
onion-platform.
It is the flagship intelligence component; the onion-platform application orchestrates it, while onion-cli can invoke its APIs directly.
MIT – see LICENSE in the root repository.