Tool Replay

What the company appears to be: Tool Replay is a self-reported local-first behavioral debugger and regression-testing system for AI applications containing non-deterministic effects. The author states it records external effects, replays them without calling real services, allows forking and modification of recorded responses, and promotes corrected behaviors into offline pytest regression tests.

What changed: The project evolved from a small experiment recording and replaying one Python function to a provider-neutral runtime supporting synchronous and sequential asynchronous applications, including OpenAI and AsyncOpenAI clients. It now supports full workflow: capture → replay → fork → diff → promote to test.

Single most important open question: Does Tool Replay have any real-world adoption or usage beyond the author's own development work?

The description states that Tool Replay is a "local-first behavioral debugger and regression-testing system for applications containing non-deterministic effects." It allows developers to:

• Record external effects
• Replay them without executing real services
• Reproduce both successful results and failures
• Fork recorded executions
• Replace exact recorded outputs
• Run applications against modified behavior
• Compare original and modified traces
• Identify first observable divergence
• Promote corrected execution into offline pytest regression case

The author describes it as a "lower-level debugging primitive" that treats AI executions like Git treats code, enabling deterministic reproduction of incidents for debugging and testing.

Evidence: The project write-up explicitly describes these capabilities in detail. It is self-reported and unverified.

The author states that Tool Replay was inspired by the need to debug AI applications where failures are difficult to reproduce due to non-determinism. The original problem was: "when an AI application failed, reproducing the exact incident was extremely difficult."

The evolution of positioning appears to be from a simple debugging tool to a complete behavioral testing framework that enables developers to create regression tests from AI incidents.

Evidence: The project write-up describes the inspiration and how the product evolved through milestones. It is self-reported and unverified.

The description states that Tool Replay is aimed at "AI developer community" and is designed as a "lower-level debugging primitive that developers could inspect, extend, run locally, and integrate into their own applications."

It targets developers working with AI models and LLM-powered workflows who need to debug non-deterministic behavior.

Evidence: The project write-up explicitly mentions targeting AI developers and the community. It is self-reported and unverified.

Not evidenced. The description does not contain any information about pricing, monetization, or business model.

The author states that Tool Replay uses:

• Python, uv, SQLite, pytest, Ruff, Pyright
• Built with asyncio, codex, gpt-5.4, gpt-5.6, openai, openai-api, python, sqlite, uv
• Supports both synchronous and asynchronous effects
• Uses SQLite for persistence, canonical JSON, SHA-256 fingerprints
• ContextVar keeps active replay session local to current execution context
• Explicit adapters for OpenAI and AsyncOpenAI
• Supports non-streaming OpenAI Response calls through explicit synchronous and asynchronous adapters

The system is described as provider-neutral, treating external operations as effects from APIs, databases, tools, or LLM providers.

Evidence: The project write-up details the technical stack and implementation choices. It is self-reported and unverified.

Not evidenced. There is no mention of customers, revenue, usage metrics, or adoption beyond the author's own development work.

The author states they looked for existing tools but found "observability platforms, evaluation frameworks, caches, and tracing systems" that did not cover the complete workflow at runtime level. They describe Tool Replay as filling a gap in this space.

Evidence: The project write-up mentions the lack of existing solutions covering the described workflow. It is self-reported and unverified.

1. No traction or adoption evidence: The tool has no demonstrated customers, usage, or revenue.
2. Single-person team: Only one member listed (Valentin Pignatelli).
3. Self-reported only: All claims are from the author's own description with no independent verification.
4. Open-source nature: While described as open source, there is no evidence of community engagement or contributions.
5. Limited scope: The project appears to be a personal development effort rather than a commercial product.

Inference: Without any evidence of real-world usage or adoption, the tool may not have sufficient market validation to warrant investment or partnership consideration.

1. What specific AI applications or workflows are you targeting with Tool Replay?
2. Have you identified any early adopters or users beyond your own development work?
3. How do you plan to monetize this tool if it's open source?
4. What is the expected timeline for broader adoption in the AI developer community?
5. Are there any technical limitations that prevent scaling beyond personal use cases?