Hey,
I’ve been working on an OSS CLI called confluence2md that exports Confluence Cloud content to local Markdown.
It starts from one or more seed pages, traverses linked pages up to a configurable depth, converts pages to Markdown, downloads referenced attachments, fetches comments, and rewrites links between crawled pages to local relative paths. It also produces a full crawl metadata graph that captures page relationships, crawl provenance, and traversal structure.
Confluence content is treated as a graph-first system and exported via a two-phase materialization pipeline, with metadata.json acting as the source-of-truth representation of structure and relationships.
I’m interested in hearing from people who have built Confluence export, migration, indexing, backup, or AI-ingestion tooling:
- What API edge cases surprised you?
- Which content types or macros tend to be painful?
- Is there metadata you wish export tools preserved?
- Are there common failure modes I should test for?