Bridge Passport

The passport explains what evidence BDC Bridge needs to evaluate an AI deployment decision. If you submit a packet without Level-1 evidence, Bridge will return untrustworthy / 0.12 / tier0 with a cautions list — that is by design, not an error. This page tells you what evidence yields a useful verdict.

What Bridge needs (Level-1 evidence)

To get an actionable verdict, attach at least these three CSV files to your packet (or in the /request-access form):

• unified_variant_comparison.csv — blank template
• current_slice_metrics.csv — blank template
• failure_case_registry.csv — blank template

For technical users calling the API directly, schemas are also available as JSON at GET /v1/passport/csv (umbrella) and GET /v1/passport/csv/<slug> (per-CSV).

What evidence yields a useful verdict

Bridge evaluates the deployment decision, not the AI capability. Useful evidence is structured, specific to your live system, and honest about what you have observed in practice. Avoid:

• Hypothetical or planned-for failure modes (must be observed)
• Aggregate-only metrics (Bridge needs slice-level decomposition)
• Benchmark-only data (your live system state matters, not leaderboard scores)
• Post-hoc rationalization (Bridge expects the structured evidence shape, not narrative)

Downloads

Bulk download all blank Level-1 templates as a single ZIP: /passport/passport-templates.zip

Source of truth

This page is a Bridge-hosted mirror of the canonical passport repository at https://github.com/malishomen/bdc-bridge-passport (pinned at v0.1.1). The Bridge mirror is for in-product discovery; the repository is canonical for schemas and examples.

Examples

Sample fully-filled packets are at /passport/examples (anonymized templates; not real partner data).

← back to bridge.bdc-hive.com