VerifAIer Sentinel · AI Governance Infrastructure · session analysis

Session Governance

Load an AI assistant conversation transcript. VerifAIer Sentinel applies deterministic session governance rules: intent drift, unsafe recommendations, warning reversals, safe-to-unsafe downgrades, and authority bypasses.

API offline. Live analysis requires the Sentinel API: python -m vai.sentinel_server or vai serve. With the API running on localhost:8000 you can paste or upload any transcript JSON for real-time governance. You can still load the built-in example below to explore a pre-analyzed session.

Session Transcript Input

Supports Claude Code, Cursor, Codex, and generic conversation formats · max 10 MB · processed locally, never uploaded

Ctrl+Enter
Running session governance pipeline…
—
governance status
0
session findings
0
turns analyzed
—
platform

Turn Timeline

Each bar = one turn   ▮ safe   ▮ questionable   ▮ unsafe   ▮ neutral

Session Findings (0)

Governance Receipt

session_id —
session_hash —
governance_status —
governance_action —
authoritative_for_wiw false — heuristic analysis only
analysis_ms —

authoritative_for_wiw=false. Session governance analysis is heuristic-based and does not constitute legal determination of who instructed what. VerifAIer Sentinel identifies behavioral patterns, not intent. All findings require human review before use in compliance or legal processes. All processing is local — no data is transmitted externally.