Python Bytes for AI
Query a decade of Python news directly from your AI assistant
Every week since 2016 we have covered what is new in Python. That is 490+ episodes of show notes, the links behind every topic, and full transcripts. All of it is now available to AI assistants through the Model Context Protocol (MCP), plain Markdown URLs for tools that do not speak MCP, and llms.txt.
MCP server
Assistants like Claude can query Python Bytes directly, with 12 tools available. No account, no API key, and nothing writable. Everything it serves is already public on this site.
- Search episode titles and show notes, or search the transcripts for what was actually said out loud.
- Read an episode in full, including its metadata, audio, and video links.
- Pull the topics out of any episode: each numbered news item, who presented it, and every link it pointed at.
- Get transcripts as plain text or timestamped WebVTT, and check when the next live stream is.
What you can ask
Once it is connected, questions like these get real answers:
- "What has Python Bytes said about uv?"
- "What packages did Python Bytes cover in episode 400?"
- "Find the Python Bytes episode where they discussed free-threaded Python."
- "Summarize the last five episodes of Python Bytes."
- "When is the next Python Bytes live stream?"
See the full MCP documentation for every tool and its parameters.
Claude Code
One command:
claude mcp add --transport http python-bytes https://pythonbytes.fm/api/mcp
Claude web (claude.ai)
- Go to Settings, then Connectors
- Click Add custom connector
- Enter the endpoint:
https://pythonbytes.fm/api/mcp
Claude Desktop and other MCP clients
Add this to your MCP configuration file, then restart the client:
{
"mcpServers": {
"python-bytes": {
"url": "https://pythonbytes.fm/api/mcp"
}
}
}
Where that file lives:
- Claude Desktop (macOS):
~/Library/Application Support/Claude/claude_desktop_config.json - Claude Desktop (Windows):
%APPDATA%\Claude\claude_desktop_config.json
Client setup moves around. If the path above does not match what you see, check your client's own documentation.
No MCP support? Add a suffix.
Plenty of AI tools still do not speak MCP, and sometimes you just want the raw text. Every episode serves alternate representations straight off its own URL, so anything that can fetch a web page can read the show.
.mdclean Markdown: title, links, show notes, transcript pointer.summarya plain-text summary.vttthe transcript with timestamps.title,.youtube,.youtube_id,.mp3for the rest
There is no separate API to learn or version. Take any episode URL and append a suffix:
https://pythonbytes.fm/episodes/show/400/celebrating-episode-400.md
Or just paste an episode link into your assistant and ask it to fetch the
.md version.
llms.txt
Following the llms.txt specification, we publish a Markdown overview of the site written for AI assistants rather than for crawlers. It documents the MCP server, every URL suffix, and the feeds, so an assistant can work out how to use Python Bytes on its own.
- Site overview at /llms.txt
- MCP discovery card at /.well-known/mcp.json
- Full RSS feed at /episodes/rss
Running python-bytes-mcp v1.1.0. Something broken, or
a tool you wish existed? Email
contact@pythonbytes.fm.
Talk Python runs
its own AI
integrations over the podcast archive and the full course catalog.