TRUST / TRANSPARENCY

Know what Vericore
does with your data.

This page describes the current 0.7.0 implementation. It is intentionally limited to behavior that exists today.

LOCAL-FIRST

Repository analysis

Vericore analyzes repositories locally. With AI disabled, deterministic repository analysis does not send repository content to a Vericore telemetry or storage service.

AI / OPT-IN

Provider boundaries

AI-assisted workflows are optional. When configured, bounded repository-derived context is sent directly to the selected provider for the requested operation. AI is not authoritative repository truth.

REPORTS

Self-contained output

Generated HTML reports contain their visualization code and do not require a browser CDN request at report-open time, allowing offline inspection.

API BOUNDARY

Trusted local use

The local REST server validates paths and applies rate limiting, but it does not provide authentication, authorization, tenant isolation, or deployment-level TLS. Do not expose it publicly without an appropriate security boundary.

PATH SAFETY

Bounded repositories

Local server and MCP boundaries reject remote repository URLs and validate repository paths. Keep allowed roots as narrow as practical when configuring trusted integrations.

CONTEXT LIMITS

Bounded AI input

AI operations use bounded repository-derived context rather than indiscriminately sending an entire repository. Exact limits and provider behavior are documented in the source repository.

CHANGE SAFETY

Verification is not authorization

The Agent Change Contract is a deterministic verification boundary. It detects mismatches, tampering, stale state, and repository identity changes; it is not an authorization system or autonomous coding mechanism.

LIMITATIONS

What is not shipped

There is no production telemetry service, multi-tenant governance layer, autonomous source modification, or deployment-grade authentication in the current release.