Would Docling be a named pilot for an open tool that records why AI-generated code changed?

[Thanatos9404]

Hi @dolfim-ibm,

Thanks for merging #4813 on JATS ext-links. Docling has both AGENTS.md and CLAUDE.md, and a case study from a project this widely used would carry real weight.

I'm applying to the Research Software Catalyst Fund for a six-month grant to build an open tool that keeps the reasoning behind AI-generated code next to the code. When someone works with Claude Code, Codex or Copilot, it saves their requests and the agent's stated reasons word for word, links them to the commit as a short Markdown record, and flags changes where one session wrote both the code and its tests. It is MIT-licensed, runs locally, and nothing leaves the machine. It builds on LLMSlim, my open-source memory library.

The fund asks who is involved, and I'd like Docling to be one of four pilot projects. If the grant is awarded (decisions by mid-March 2027), a pilot means:

• trying the tool on real work for four weeks or more in July and August 2027,
• one 45-minute interview, and
• reviewing a short case study before it is published under CC BY.

Each pilot project receives a US$1,000 honorarium. A yes now commits you to nothing unless the grant comes through, and you can drop out at any time.

The proposal is submitted, and I can update it until 31 December. May I name Docling as a prospective pilot? A one-line reply here by 15 December is plenty, and "no thanks" is completely fine.

Thanks,
Yashvardhan Thanvi
github.com/Thanatos9404