Repository navigation
Home en
ClarusIubar edited this page Aug 17, 2026
·
5 revisions
한국어 | English
A unified pipeline that converts ChatGPT / Gemini / Claude takeout exports
into Obsidian-compatible markdown. Each vendor's raw export format is
completely different, but shared logic — code-fence safety, frontmatter
assembly, callout formatting — lives in common/ and is shared across
vendors.
The canonical source is README.md at the repository root; this wiki is a projection that reorganizes that content by purpose for easier reading. For the latest details, always check the README and the code.
-
Getting Started — laying out the
data/folder, runningrun.py,--input/--publish -
Configuration — the three
config.jsonpaths, CLI override priority, exit codes - Output Format — the generated markdown structure, turn-metadata HTML comments
-
Architecture — how
common//vendors//run.pyare organized - Development — running the tests, pure-function design
-
MCP Experiment Research Log — the
mcp-experiment-without-RAGbranch, a stage-by-stage record of what was tested, why, and what did/didn't work (main is frozen as stablev0.1.0; this branch isn't merged there)
| What you're wondering | Where to go |
|---|---|
| How do I run this the first time | Getting Started |
| How do I point it at my takeout/output/vault paths | Configuration |
| Why is the generated note shaped this way, how does RAG chunking work | Output Format |
| How is the code split up | Architecture |
| How do I run the tests | Development |
| Report a security vulnerability | SECURITY.md |