Skip to content

About ai-blame

Project

ai-blame is a tool for extracting and tracking provenance from AI agent execution traces. It emerged from work on AI-assisted curation of knowledge bases, ontologies, and structured data, but applies broadly to any code or content generated with AI assistance.

Origin

This project is part of the broader AI4Curation ecosystem—a collection of patterns, tools, and workflows for maintaining knowledge bases and code in an age of AI pair programming.

Key Use Cases

  • Knowledge base curation — Track which AI models generated or edited ontology files
  • Code auditing — Understand which parts of your codebase are AI-generated
  • Compliance — Create audit trails for industries with strict provenance requirements
  • Model evaluation — Compare output quality across different model versions

Read the Article

For a comprehensive introduction to the problem and solution, see:

"Whose Code Is This, Anyway? Tracking AI Agent Provenance" — A Medium article exploring: - Why git blame isn't enough in AI-assisted workflows - Real-world examples from knowledge base curation - How embedded provenance solves attribution and accountability gaps - The regulatory and compliance implications

License

BSD-3-Clause

Contributing

Contributions welcome! See the main repository for ways to contribute.