Agent Audit
Reads a JSON or JSONL agent log and reports what was proposed, approved, refused, completed, and failed.
Type
Tech stack
Documentation
Basic
A README with setup and run steps.
Maintenance
None
Sold as-is. No further updates from the Builder.
Description
Agent Audit — Execution Logs Into Readable Reports
Agent Audit is a local Python utility that reads AI-agent execution logs and turns them into a clear, structured report. Events are grouped as proposed, approved, refused, completed, or failed, with timestamps and recorded reasons wherever the log provides them.
Built from the Voxryn FREE runtime and sold separately, Agent Audit works as a standalone log reader. It does not run the agent, enforce permissions, or invent missing information.
Unpack the zip and run:
python agent_audit.py sample/agent.jsonl
A flat record uses at, action, status, and reason. A FREE constitutional audit line uses ts, event_type, and payload. When payload.allowed is false, the row is reported as refused. A row that fits neither shape is listed under other, so it is not dropped quietly.
It reads the log. It does not run an agent. It does not need the FREE runtime.
The zip is the source, a sample log, and tests. Python 3.12, standard library only. No install. Copyright 2026 Martin Gettig. No support, hosting, or updates. This license is for Agent Audit. It is not the FREE runtime.
FAQ
- Does this run my agent?
- No. It reads a JSON or JSONL log you already have and prints a report.
- Do I need the FREE runtime?
- No. The tool runs on its own. One of the log shapes it reads is a FREE constitutional audit line.
Seller
Reviews
No reviews yet.