Structured Evidence
Every orchestration run produces a complete audit trail: claims made, proofs collected, verdicts rendered. Full accountability for AI-generated code.
What gets tracked
Evidence Ledger
- • Claims: What each step promises to accomplish
- • Proofs: Command output, file diffs, verification results
- • Verdicts: PASS, FAIL, or INFO_NEEDED with reasoning
Interaction History
- • Full LLM prompts and responses
- • Token usage (prompt_eval_count, eval_count)
- • Session grouping for multi-turn interactions
Event Log
- • Structured events for every orchestration action
- • Status transitions, phase changes, error traces
- • Filterable by type, request, and time range
Screenshots
- • Automated capture for visual verification
- • Configurable viewport and full-page options
- • Disk usage tracking and cleanup
See it in the dashboard
Event Log
Interaction History