Matthew0822/ToolReplay
Audit AI agent tool-call transcripts: hash-chain sealing, deterministic replay, and scope overreach checks. Dependency-free Python CLI.
Older snapshot: check the original source for current information.
From the publisher
README & documentation
Earlier source previewRead the project’s overview, installation instructions and usage examples. The original README is the source of truth.
ToolReplay audits a recorded transcript of an AI agent's tool calls and reports where the session was non-deterministic, redundant, or outside its declared permissions. Here is the tool working, right now, on the dirty sample that ships in this repository. The command is shown above its output, and the output is pasted verbatim from a run in this checkout: Those three header lines and two findings are the whole audit of a six-call session. Read them top to bottom. calls: 6 is the number of records parsed from samples/session-dirty.jsonl. Parsing is strict, so a six here means six well-formed lines with indices 0 through 5 and nothing malformed. divergence: index 5 is the first index where a deterministic re-run would disagree with the record. The tool found that index 5 repeated an earlier call but carried a different recorded response, so that is the earliest point the session stops being…
The latest refresh did not succeed. This is the last available preview; check the original for changes.
Read the full README ↗ · Preview checked 2026-09-30T14:55:18.825Z
Inside the original README — Document outline
- ToolReplay
- What just happened
- Install
- Commands
- The three findings it detects
- Non-determinism
- Redundant call
- Permission overreach
- The transcript format
- Sealing and the hash chain
- Scope declaration format
- Output format
Headings are captured from the source. Links open the publisher’s document, not a locally hosted copy.
Links from the README
References supplied by the publisher, not independently verified endorsements. Check the destination before downloading files or entering credentials.
Documentation belongs to its respective authors. Reported project/model license: MIT. A listing is not a grant of reuse or training rights. Confirm the document’s own terms at the source.
What this repository does
Audit AI agent tool-call transcripts: hash-chain sealing, deterministic replay, and scope overreach checks. Dependency-free Python CLI.
Repository facts
- Owner
- Matthew0822
- Primary language
- Python
- Stars
- 180
- Forks
- 21
- Open issues + pull requests
- 1
- License
- MIT
- Archived
- No
- Default branch
- main
- Created
- 2026-09-14T10:11:00.000Z
- Last push
- 2026-09-14T12:09:08.000Z
Topics and intended use
Owner-supplied topics: ai-agents, audit, cli, determinism, hash-chain, jsonl, permissions, python, replay, scope, security, tool-call
Review the README for scope, installation, examples and limitations. We do not run repository code or certify it.
Evaluate before installing
Review licensing and dependencies, inspect recent commits and unresolved issues, and test in an isolated environment before production use. Stars and forks alone cannot answer these questions.
Source and freshness
Source: GitHub. Metadata observed 2026-09-20T06:40:45.138Z. Daily imports are snapshots, not real-time monitoring.
Popularity and source listings do not establish security, suitability, licensing rights or benchmark performance.
Related repositories
EbookFoundation/free-programming-books
donnemartin/system-design-primer