Keep code blocks, language tags, and Claude artifacts intact
Most exporters treat code like body text. ChatExport AI keeps language tags (```python stays ```python), preserves indentation byte-perfect, extracts Claude artifacts as standalone files in a ZIP, and emits JSON ready for RAG ingestion or fine-tunes.
Code stays code. Artifacts stay files.
Language tags preserved
```ts in equals ```ts out. Markdown exports keep fenced blocks with the language hint intact for syntax highlighting downstream.
Claude artifacts as files
HTML artifacts, React components, SVG, Mermaid — extracted as separate files in a ZIP, referenced from the transcript by relative path. Real review-ready review-ready output.
Byte-perfect indentation
Tabs vs spaces, four vs two — exactly as the AI emitted. No re-indenting in your editor.
JSON with role / content / metadata
Each turn structured: role, content, model, timestamp. Drop into a RAG pipeline or fine-tune dataset.
From chat to commit in three steps
Pair with Claude / ChatGPT / Cursor
Ship a feature, refactor, debug — whatever. The conversation has reasoning that's worth keeping.
Export Markdown + JSON
Markdown for the docs commit, JSON for your retrieval pipeline.
Diff-friendly artifacts in git
Markdown sits cleanly next to the code, reviewable in PRs, diffable across model iterations.
Pick what your toolchain wants
Code-faithful AI exports
Free Chrome extension. Language tags. Artifacts. JSON.
Add to Chrome — It's Free