| Name | claude2md JSON |
| Version |
0.1
JSON |
| download |
| home_page | None |
| Summary | Convert Claude.ai or Claude Code chats to Markdown |
| upload_time | 2025-10-26 09:17:23 |
| maintainer | None |
| docs_url | None |
| author | None |
| requires_python | >=3.10 |
| license | None |
| keywords |
claude
markdown
chat
export
converter
|
| VCS |
 |
| bugtrack_url |
|
| requirements |
No requirements were recorded.
|
| Travis-CI |
No Travis.
|
| coveralls test coverage |
No coveralls.
|
# `claude2md`
Convert Claude.ai or Claude Code chat exports to Markdown format.
`claude2md` converts chats exported from Claude.ai or Claude Code to Markdown format. This lets you efficiently show entire chats to other models or instances until you struggle to remember which transcript of a chat full of transcripts of chats full of transcripts of chats is the right one (aka slopception).
## Features
You can filter the exports to any subset of user, assistant, and thinking blocks. For context management I've found it's often helpful to only export user messages, though obviously for the strongest "continuity" between an exported chat transcript and its consumer you want all context from the prior chat. Though Claude typically does not get to see its past thinking, so `--no-thinking` (our default) ought to be just as good if not better than including thinking blocks (by virtue of being on-distribution).
We try to include Claude.ai attachments when they're available in the export, but the format we parse right now does not have them embedded in the JSON or anything, so in practice only attachments with `extracted_content`, i.e. text files, markdown, etc. will be present in the transcripts.
You can branch conversations on Claude.ai (and to a lesser extent in Claude Code), and `claude2md` can export any branch. You can list the branches of a chat and their UUIDs with `claude2md --branches` and export a specific branch with `claude2md --branch UUID`.
We don't support "Claude Code on the Web" yet, though ostensibly if you "teleport" a Web session to a local Claude Code instance, you can export the history as normal from there.
## Install
The blazing fast and memory safe way:
$ uvx claude2md --help # installed on demand
The traditional way:
$ pip install claude2md
$ claude2md --help
The bleeding-edge way:
$ git clone https://github.com/twilligon/claude2md
$ cd claude2md
$ python3 -m venv venv; . venv/bin/activate # you probably want a venv
$ pip install -e .
$ claude2md --help
## Usage
$ claude2md --help
usage: claude2md [-h] [--branches | --branch UUID] [--user] [--no-user]
[--assistant] [--no-assistant] [--thinking]
[--no-thinking] [--title] [--no-title]
[file]
Convert Claude.ai or Claude Code chats to Markdown
positional arguments:
file JSON file to convert (default: stdin)
options:
-h, --help show this help message and exit
--branches List all branch message UUIDs
--branch UUID Show chain to specific branch
--user Show user messages (default)
--no-user Hide user messages
--assistant Show assistant (default)
--no-assistant Hide assistant
--thinking Show thinking blocks
--no-thinking Hide thinking (default)
--title Always show title (or "Untitled" if none)
--no-title Never show title
## Examples
#### Claude.ai
$ claude2md --title --thinking ~/.local/share/claude-backup/f50d0792-fa79-4d2d-a945-c8f709e9e7a7/c6a33008-39b3-45ed-a00e-413141c89a90.json | head -n20
# Python packaging complexity
> why is python packaging such a clusterfuck? ultrathink
<thinking>
The user is asking about Python packaging and why it's so complicated/messy. They've also said "ultrathink" which I interpret as wanting a very thorough, deep analysis.
This is a great question that many Python developers have strong feelings about. Let me think through the historical and technical reasons why Python packaging became so complex.
Key issues to cover:
1. Historical evolution - how we got here
2. Multiple competing tools and standards
3. The various problems each tool tries to solve
4. Why there's no single "right" way
5. The current state and attempts to fix things
Let me structure this thoughtfully and comprehensively since they asked for "ultrathink."
</thinking>
Python packaging *is* notoriously complex, and there are deep historical and technical reasons for this. Let me break down why it became such a mess:
#### Claude Code
$ claude2md ~/.claude/projects/-home-claude-claude2md/0c8de429-fb29-4096-925e-6d81a0437f09.jsonl --branch 670e0906-013b-4c83-b7c7-ecf8e12aab3e | head -n3
> rename --leaves and --leaf to --branches and --branch. no need to change how we refer to this stuff internally, only user facing matters. in @claude2md/__init__.py . ultrathink
I'll help you rename the command-line arguments from `--leaves`/`--leaf` to `--branches`/`--branch` in the user-facing interface. Let me start by creating a todo list for this task.
## License
`claude2md` is dedicated to the public domain where possible via CC0-1.0.
Raw data
{
"_id": null,
"home_page": null,
"name": "claude2md",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.10",
"maintainer_email": null,
"keywords": "claude, markdown, chat, export, converter",
"author": null,
"author_email": "Ultrathinkers Anonymous <ultrathink@twilligon.com>",
"download_url": "https://files.pythonhosted.org/packages/22/d5/ce792250c50bf8e830e57219341d6197c2fe2f5ef48936698cb63b29d4a6/claude2md-0.1.tar.gz",
"platform": null,
"description": "# `claude2md`\n\nConvert Claude.ai or Claude Code chat exports to Markdown format.\n\n`claude2md` converts chats exported from Claude.ai or Claude Code to Markdown format. This lets you efficiently show entire chats to other models or instances until you struggle to remember which transcript of a chat full of transcripts of chats full of transcripts of chats is the right one (aka slopception).\n\n## Features\n\nYou can filter the exports to any subset of user, assistant, and thinking blocks. For context management I've found it's often helpful to only export user messages, though obviously for the strongest \"continuity\" between an exported chat transcript and its consumer you want all context from the prior chat. Though Claude typically does not get to see its past thinking, so `--no-thinking` (our default) ought to be just as good if not better than including thinking blocks (by virtue of being on-distribution).\n\nWe try to include Claude.ai attachments when they're available in the export, but the format we parse right now does not have them embedded in the JSON or anything, so in practice only attachments with `extracted_content`, i.e. text files, markdown, etc. will be present in the transcripts.\n\nYou can branch conversations on Claude.ai (and to a lesser extent in Claude Code), and `claude2md` can export any branch. You can list the branches of a chat and their UUIDs with `claude2md --branches` and export a specific branch with `claude2md --branch UUID`.\n\nWe don't support \"Claude Code on the Web\" yet, though ostensibly if you \"teleport\" a Web session to a local Claude Code instance, you can export the history as normal from there.\n\n## Install\n\nThe blazing fast and memory safe way:\n\n $ uvx claude2md --help # installed on demand\n\nThe traditional way:\n\n $ pip install claude2md\n $ claude2md --help\n\nThe bleeding-edge way:\n\n $ git clone https://github.com/twilligon/claude2md\n $ cd claude2md\n $ python3 -m venv venv; . venv/bin/activate # you probably want a venv\n $ pip install -e .\n $ claude2md --help\n\n## Usage\n\n $ claude2md --help\n usage: claude2md [-h] [--branches | --branch UUID] [--user] [--no-user]\n [--assistant] [--no-assistant] [--thinking]\n [--no-thinking] [--title] [--no-title]\n [file]\n\n Convert Claude.ai or Claude Code chats to Markdown\n\n positional arguments:\n file JSON file to convert (default: stdin)\n\n options:\n -h, --help show this help message and exit\n --branches List all branch message UUIDs\n --branch UUID Show chain to specific branch\n --user Show user messages (default)\n --no-user Hide user messages\n --assistant Show assistant (default)\n --no-assistant Hide assistant\n --thinking Show thinking blocks\n --no-thinking Hide thinking (default)\n --title Always show title (or \"Untitled\" if none)\n --no-title Never show title\n\n## Examples\n\n#### Claude.ai\n\n $ claude2md --title --thinking ~/.local/share/claude-backup/f50d0792-fa79-4d2d-a945-c8f709e9e7a7/c6a33008-39b3-45ed-a00e-413141c89a90.json | head -n20\n # Python packaging complexity\n\n > why is python packaging such a clusterfuck? ultrathink\n\n <thinking>\n The user is asking about Python packaging and why it's so complicated/messy. They've also said \"ultrathink\" which I interpret as wanting a very thorough, deep analysis.\n\n This is a great question that many Python developers have strong feelings about. Let me think through the historical and technical reasons why Python packaging became so complex.\n\n Key issues to cover:\n 1. Historical evolution - how we got here\n 2. Multiple competing tools and standards\n 3. The various problems each tool tries to solve\n 4. Why there's no single \"right\" way\n 5. The current state and attempts to fix things\n\n Let me structure this thoughtfully and comprehensively since they asked for \"ultrathink.\"\n </thinking>\n\n Python packaging *is* notoriously complex, and there are deep historical and technical reasons for this. Let me break down why it became such a mess:\n\n#### Claude Code\n\n $ claude2md ~/.claude/projects/-home-claude-claude2md/0c8de429-fb29-4096-925e-6d81a0437f09.jsonl --branch 670e0906-013b-4c83-b7c7-ecf8e12aab3e | head -n3\n > rename --leaves and --leaf to --branches and --branch. no need to change how we refer to this stuff internally, only user facing matters. in @claude2md/__init__.py . ultrathink\n\n I'll help you rename the command-line arguments from `--leaves`/`--leaf` to `--branches`/`--branch` in the user-facing interface. Let me start by creating a todo list for this task.\n\n## License\n\n`claude2md` is dedicated to the public domain where possible via CC0-1.0.\n\n",
"bugtrack_url": null,
"license": null,
"summary": "Convert Claude.ai or Claude Code chats to Markdown",
"version": "0.1",
"project_urls": {
"Issues": "https://github.com/twilliddgon/claude2md/issues",
"Repository": "https://github.com/twilligon/claude2md"
},
"split_keywords": [
"claude",
" markdown",
" chat",
" export",
" converter"
],
"urls": [
{
"comment_text": null,
"digests": {
"blake2b_256": "00e2e69d53370f0bd9550dc87352dee51d43522ce8e7730788cb3c4b8893a611",
"md5": "708c7a8336a104fabb358fedece3efb6",
"sha256": "3a24108e67c3f9c6ff6a9692642d4236e035d95cd6fa4537e875762f554d583a"
},
"downloads": -1,
"filename": "claude2md-0.1-py3-none-any.whl",
"has_sig": false,
"md5_digest": "708c7a8336a104fabb358fedece3efb6",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.10",
"size": 9720,
"upload_time": "2025-10-26T09:17:22",
"upload_time_iso_8601": "2025-10-26T09:17:22.735242Z",
"url": "https://files.pythonhosted.org/packages/00/e2/e69d53370f0bd9550dc87352dee51d43522ce8e7730788cb3c4b8893a611/claude2md-0.1-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": null,
"digests": {
"blake2b_256": "22d5ce792250c50bf8e830e57219341d6197c2fe2f5ef48936698cb63b29d4a6",
"md5": "a1ccb654b5459d3724d2c74b8d949054",
"sha256": "1564d0a908754d0437c5c07257d41aa14429788eda9d41bb8887b9c1e2e55801"
},
"downloads": -1,
"filename": "claude2md-0.1.tar.gz",
"has_sig": false,
"md5_digest": "a1ccb654b5459d3724d2c74b8d949054",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.10",
"size": 8697,
"upload_time": "2025-10-26T09:17:23",
"upload_time_iso_8601": "2025-10-26T09:17:23.902043Z",
"url": "https://files.pythonhosted.org/packages/22/d5/ce792250c50bf8e830e57219341d6197c2fe2f5ef48936698cb63b29d4a6/claude2md-0.1.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2025-10-26 09:17:23",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "twilliddgon",
"github_project": "claude2md",
"github_not_found": true,
"lcname": "claude2md"
}
