Should your agent call an MCP server or just run a CLI? Pamela Fox is a principal cloud advocate at Microsoft and GitHub. She’s taught MCP workshops from PyCon to WeAreDevelopers. Every week, an agent combs her Gmail, Outlook, and X history to build her news roundup. She uses MCP all day. She’ll still tell you when to skip it. Her colleague pitted the GitHub CLI against the GitHub MCP server on a real bug fix. The CLI used fewer tokens. It succeeded more often. So do we choose CLIs, MCPs, or even just APIs for connecting our agents to our data and services?

NumPy already releases the GIL. So threads should already pay off for your NumPy code, right? Christopher Ariza tested that. He summed the rows of a 10,000 by 10,000 array on Python 3.14. Sequential took 360 milliseconds. A thread pool took 373, slower than doing nothing. The same code on free-threaded Python 3.14t took 83.7 milliseconds. Free-threaded Python is officially supported as of 3.14. The GIL isn’t gone, though. It’s a zombie. Import one compiled package that hasn’t opted in, psycopg2 for example, and the GIL comes back for your entire process. You get a warning, not an error.

Chris is the creator of StaticFrame and its C extension library ArrayKit. We dig into when threads beat plain NumPy, how to tell if you’re actually running free-threaded, and conditional-futures, his drop-in thread pool that stops your code from getting slower when the GIL wakes back up.

"The podcast has been a wealth of information and has positively contributed to my efforts to learn Python. Frankly, the podcast is the best way I've found to augment the training videos, documentation, and books I've used to learn the language."

"This is the podcast that got me into listening to nearly only technical ones. All the episodes are fun and informative, and I can't recommend it enough for every Pythonista."

"Each and every episode, I learn at least one thing that motivates me to change my day to day life."

MCP Server

Connect Claude, Cursor, or any MCP client to talkpython.fm/api/mcp and get 12 tools for searching episodes, reading transcripts, and browsing courses.

https://talkpython.fm/api/mcp

Command Line

No MCP support in your AI tool? The Talk Python CLI does the same job from a terminal, with text, JSON, or Markdown output ready to pipe into anything.

uv tool install talk-python-cli
talkpython episodes search "FastAPI"

Python Bytes Podcast

Do you enjoy Talk Python To Me? We created a second podcast to be the perfect counterpart to the long-form interview format.

Python Bytes delivers Python headlines directly to your earbuds in just 15-20 minutes per week. The topical items of the Python ecosystem, condensed and ready to go.

About the Show

Talk Python to Me is a weekly podcast for Python developers, data scientists, and AI builders. Each episode features a casual 1-hour conversation with the people building the tools you use.

Have feedback for the show? We'd love to hear from you.

Suggest a show or guest

We are always on the look out for exciting topics to bring to you on Talk Python To Me. Do you have a great idea for a show? Know a dynamic Python personality who we should be speaking to?