Idea: “Phoenix docs → RAG-ready Markdown” export (clean ingestion for internal assistants)

[HEDELKA]

Hi Phoenix team!

I’m working on a documentation scraper for RAG pipelines: it converts docs sites into clean Markdown (removes nav/footers, preserves code blocks), and adds metadata + optional chunking so it can be directly loaded into vector DBs.

I think this could be useful for Phoenix users who:

• want an internal “chat with Phoenix docs” assistant
• need a reproducible, scriptable docs export for evaluation/experimentation workflows
• want to keep an offline/archived copy of docs in a structured format

Questions:

1. Would you be open to a community recipe like “Export Phoenix docs to an LLM/RAG corpus”?
2. Is there a preferred place to publish it (Docs, Discussions, separate repo)?
3. Any requirements around attribution/licensing for distributing the exported Markdown dataset?

If you want, I can share a small sample export (a couple pages) so you can judge output quality before we proceed.

[mikeldking]

hey @HEDELKA we just switched to Mintlify which has markdown rendering of all our pages and it also has MCP.

In general we're always open to PRs but we def. end up needing to maintain the packages so depending on it's utility we might be a bit picky given the amount of code we have to maintain.

Hope that helps.

[HEDELKA]

Thanks for the heads-up on Mintlify! That definitely makes raw access easier.

I totally get the concern about maintenance burden. Instead of a code PR to the core repo, I could just share a standalone Colab Notebook / Gist that shows a "recipe" for ingestion.

It would basically demonstrate: "How to pull Phoenix docs -> Clean/Chunk -> Load into Vector Store" using external tools. This way, there is zero code for you to maintain, just a useful resource/link for the community.

Would a link to such an external guide be welcome in a "Community Resources" section or Discussions?