Artifact system
The artifacts are the product contract.
Ayneye becomes useful when the output is not a vague summary but a set of files and endpoints developers can inspect, validate, and reuse.
Artifact inventory
| Artifact | Role | Primary consumer |
|---|---|---|
| capture_report.json | What was processed and under what bounds | Operations and debugging |
| detector_report.json | Detector mode, confidence, fallback, limitations | Engineering and QA |
| timeline.json | Ordered events and semantic refresh points | Dashboard and search |
| scene_graph.json | Objects, zones, relationships, evidence_refs | APIs and agents |
| evidence_bundle.json | Traceable evidence spans and answer support | Reviewers and Ask Video |
| cost.json | Duration, mode, fallback, retention, cost context | Finance and product |
Artifact handling rules
- Treat artifacts as versioned product outputs.
- Do not let agents answer without artifact references.
- Do not hide detector uncertainty from users.
- Preserve evidence refs in UI and API responses.
- Surface cost context next to repeated-query workflows.
- Use retention limits and tenant boundaries consistently.
Product boundaries are part of the product, not footnotes. Ayneye is not presented as a replacement for every GPU video foundation model, not a free-infinite-query engine, and not an autonomous surveillance decision system. The current beta path is controlled signup, dashboard, REST API, bounded processing, evidence artifacts, visible limits, and read-only agent/MCP-style evaluation. Hard scenes, identity-sensitive workflows, emergency response, physical access, discipline, and destructive actions require human review or remain blocked.