Zeke Sikelianos
@ziki.boo
Aging Twitter user from 2007, back when it was a service where you texted messages to 40404 and they blasted to all your followers. Now scared of Elon and X, and optimistic about this AT PROTOCOL thing.
Anyone doing GitHub MCP stuff in Cursor? I'm thinking something like:
> Read the feedback I got on <PR_URL> and fix it.
☝️ Working in a private repo, so this would need authenticated access to GitHub
> Read the feedback I got on <PR_URL> and fix it.
☝️ Working in a private repo, so this would need authenticated access to GitHub
May 14, 2025 at 7:41 PM
Anyone doing GitHub MCP stuff in Cursor? I'm thinking something like:
> Read the feedback I got on <PR_URL> and fix it.
☝️ Working in a private repo, so this would need authenticated access to GitHub
> Read the feedback I got on <PR_URL> and fix it.
☝️ Working in a private repo, so this would need authenticated access to GitHub
I think Google AI Studio is amazing. Why does it get so much shade?
May 7, 2025 at 3:02 AM
I think Google AI Studio is amazing. Why does it get so much shade?
Lazyweb of AI devs: What are your favorite Node.js and Python tools for stuffing files from disk into an LLM-friendly string?
For example: github.com/simonw/file...
For example: github.com/simonw/file...
GitHub - simonw/files-to-prompt: Concatenate a directory full of files into a single prompt for use with LLMs
Concatenate a directory full of files into a single prompt for use with LLMs - simonw/files-to-prompt
github.com
March 11, 2025 at 5:14 PM
Lazyweb of AI devs: What are your favorite Node.js and Python tools for stuffing files from disk into an LLM-friendly string?
For example: github.com/simonw/file...
For example: github.com/simonw/file...
It’s true. It works so well. We’ve finally moved beyond the whisper era.
🔦 PSA:
Gemini 2.0 can accurately transcribe - with speaker identities and timestamps! - both audio and video. Just give it the file, and ask for speaker identities (if known) or generic (if not).
With Flash-Lite, transcription costs only ~$0.01 for this ~1min long commercial:
Gemini 2.0 can accurately transcribe - with speaker identities and timestamps! - both audio and video. Just give it the file, and ask for speaker identities (if known) or generic (if not).
With Flash-Lite, transcription costs only ~$0.01 for this ~1min long commercial:
March 3, 2025 at 10:07 PM
It’s true. It works so well. We’ve finally moved beyond the whisper era.
Reposted by Zeke Sikelianos
I really feel like fashion retail websites should let you browse in latent space. Don't select a category like "shirt". Instead, see some product images, and select one that's a shirt. Then see a grid of shirts, and pick your favorites. See a grid of shirts like these. Repeat.
February 24, 2025 at 2:24 AM
I really feel like fashion retail websites should let you browse in latent space. Don't select a category like "shirt". Instead, see some product images, and select one that's a shirt. Then see a grid of shirts, and pick your favorites. See a grid of shirts like these. Repeat.
X was all I knew,
Blue Sky hums a softer tune,
will I learn the words?
Blue Sky hums a softer tune,
will I learn the words?
February 23, 2025 at 6:40 AM
X was all I knew,
Blue Sky hums a softer tune,
will I learn the words?
Blue Sky hums a softer tune,
will I learn the words?
@hipstersmoothie.com thanks for building www.bluesky-migrate.com -- it's helping me get my act together to switch. 🙏
Bluesky Migrate
This page serves a simple guide on how to migrate to Bluesky from X. All steps are optional, but you should really do the first two.
www.bluesky-migrate.com
February 23, 2025 at 6:33 AM
@hipstersmoothie.com thanks for building www.bluesky-migrate.com -- it's helping me get my act together to switch. 🙏
; @replicate shipped some nice little docs updates and API improvements this week.
https://replicate.com/changelog/2025-02-14-the-little-things-week-ending-february-14-2025
https://replicate.com/changelog/2025-02-14-the-little-things-week-ending-february-14-2025
The little things, week ending February 14, 2025
A collection of updates to Replicate for the week ending ...
replicate.com
February 23, 2025 at 11:04 AM
; @replicate shipped some nice little docs updates and API improvements this week.
https://replicate.com/changelog/2025-02-14-the-little-things-week-ending-february-14-2025
https://replicate.com/changelog/2025-02-14-the-little-things-week-ending-february-14-2025
@LeonardoAi_ hitting a TOS bug at signup on iOS…
February 23, 2025 at 11:04 AM
@LeonardoAi_ hitting a TOS bug at signup on iOS…
Google's latest text-to-image model is now on @replicate, and it's really good.
It's also really easy to run from the web or the API, just like any other Replicate model. No need to fuss around with Google APIs to get up and running.
https://replicate.com/google/imagen-3
It's also really easy to run from the web or the API, just like any other Replicate model. No need to fuss around with Google APIs to get up and running.
https://replicate.com/google/imagen-3
google/imagen-3 – Run with an API on Replicate
Google's highest quality text-to-image model, capable of ...
replicate.com
February 23, 2025 at 11:04 AM
Google's latest text-to-image model is now on @replicate, and it's really good.
It's also really easy to run from the web or the API, just like any other Replicate model. No need to fuss around with Google APIs to get up and running.
https://replicate.com/google/imagen-3
It's also really easy to run from the web or the API, just like any other Replicate model. No need to fuss around with Google APIs to get up and running.
https://replicate.com/google/imagen-3
Create and Multistream Live Video | Restream
Restream is a simple live video platform that empowers yo...
Restream.io
February 23, 2025 at 11:04 AM
@craigsdennis your awesome talk-to-javascript-openai-workers stopped working! 😿
I was just gonna turn it into a proper OpenAI + CloudFlare + Replicate guide, but it looks the Realtime WebRTC API beta has drifted a bit in the last month or so. Do you have any plans to revisit
I was just gonna turn it into a proper OpenAI + CloudFlare + Replicate guide, but it looks the Realtime WebRTC API beta has drifted a bit in the last month or so. Do you have any plans to revisit
February 23, 2025 at 11:04 AM
@craigsdennis your awesome talk-to-javascript-openai-workers stopped working! 😿
I was just gonna turn it into a proper OpenAI + CloudFlare + Replicate guide, but it looks the Realtime WebRTC API beta has drifted a bit in the last month or so. Do you have any plans to revisit
I was just gonna turn it into a proper OpenAI + CloudFlare + Replicate guide, but it looks the Realtime WebRTC API beta has drifted a bit in the last month or so. Do you have any plans to revisit
@jeremyphoward @HamelHusain I love llms.txt and want to help spread the gospel.
What tools do you know of that have built-in support for it?
Like... user mentions http://example.com and the tool knows to go looking for http://example.com/llms.txt
What tools do you know of that have built-in support for it?
Like... user mentions http://example.com and the tool knows to go looking for http://example.com/llms.txt
Example Domain
example.com
February 23, 2025 at 11:04 AM
@jeremyphoward @HamelHusain I love llms.txt and want to help spread the gospel.
What tools do you know of that have built-in support for it?
Like... user mentions http://example.com and the tool knows to go looking for http://example.com/llms.txt
What tools do you know of that have built-in support for it?
Like... user mentions http://example.com and the tool knows to go looking for http://example.com/llms.txt
You know what's great? Documentation.
You know what's even better? Not having to read it.
Here's me feeding https://cog.run/llms.txt into @cursor_ai chat to help me create a @replicate model:
You know what's even better? Not having to read it.
Here's me feeding https://cog.run/llms.txt into @cursor_ai chat to help me create a @replicate model:
February 23, 2025 at 11:09 AM
You know what's great? Documentation.
You know what's even better? Not having to read it.
Here's me feeding https://cog.run/llms.txt into @cursor_ai chat to help me create a @replicate model:
You know what's even better? Not having to read it.
Here's me feeding https://cog.run/llms.txt into @cursor_ai chat to help me create a @replicate model:
Building @replicate models with Cog?
Here's a cheatsheet for your AI code editor: https://cog.run/llms.txt
Here's a cheatsheet for your AI code editor: https://cog.run/llms.txt
February 23, 2025 at 11:04 AM
Building @replicate models with Cog?
Here's a cheatsheet for your AI code editor: https://cog.run/llms.txt
Here's a cheatsheet for your AI code editor: https://cog.run/llms.txt
Having trouble getting the dog to flip the table.
February 23, 2025 at 11:09 AM
Having trouble getting the dog to flip the table.
The `all-the-public-replicate-models` npm package now has daily run counts for all the public models on @replicate.
Somebody build a leaderboard! (Looking at you, @aimodelsfyi)
Somebody build a leaderboard! (Looking at you, @aimodelsfyi)
February 23, 2025 at 11:09 AM
The `all-the-public-replicate-models` npm package now has daily run counts for all the public models on @replicate.
Somebody build a leaderboard! (Looking at you, @aimodelsfyi)
Somebody build a leaderboard! (Looking at you, @aimodelsfyi)
@ellipsis_dev nice to meet you :)
February 23, 2025 at 11:09 AM
@ellipsis_dev nice to meet you :)
The @LumaLabsAI "ray" model on @replicate (aka Dream Machine) lets you generate video using a text prompt, a starting frame, and an ending frame.
Pretty wild.
Pretty wild.
February 23, 2025 at 11:10 AM
The @LumaLabsAI "ray" model on @replicate (aka Dream Machine) lets you generate video using a text prompt, a starting frame, and an ending frame.
Pretty wild.
Pretty wild.
The @replicate playground supports video models!
1. Generate an image (e.g. with your Flux fine-tune)
2. Use it as a starting frame for a video
3. Kick off several generations at once
4. Kick back and enjoy the show
Links in next post...
1. Generate an image (e.g. with your Flux fine-tune)
2. Use it as a starting frame for a video
3. Kick off several generations at once
4. Kick back and enjoy the show
Links in next post...
February 23, 2025 at 11:10 AM
The @replicate playground supports video models!
1. Generate an image (e.g. with your Flux fine-tune)
2. Use it as a starting frame for a video
3. Kick off several generations at once
4. Kick back and enjoy the show
Links in next post...
1. Generate an image (e.g. with your Flux fine-tune)
2. Use it as a starting frame for a video
3. Kick off several generations at once
4. Kick back and enjoy the show
Links in next post...
I realize "800x600" probably means nothing to anyone under the age of 30. Showing my age here...
Are you using @replicate's API to generate images?
If you wanna keep those images for longer than an hour, you'll need to store them yourself. @cloudfare workers are a perfect fit for that.
I wrote a guide to get you going. 🧵
If you wanna keep those images for longer than an hour, you'll need to store them yourself. @cloudfare workers are a perfect fit for that.
I wrote a guide to get you going. 🧵
February 23, 2025 at 11:15 AM
I realize "800x600" probably means nothing to anyone under the age of 30. Showing my age here...
Are you using @replicate's API to generate images?
If you wanna keep those images for longer than an hour, you'll need to store them yourself. @cloudfare workers are a perfect fit for that.
I wrote a guide to get you going. 🧵
If you wanna keep those images for longer than an hour, you'll need to store them yourself. @cloudfare workers are a perfect fit for that.
I wrote a guide to get you going. 🧵
February 23, 2025 at 11:10 AM
Are you using @replicate's API to generate images?
If you wanna keep those images for longer than an hour, you'll need to store them yourself. @cloudfare workers are a perfect fit for that.
I wrote a guide to get you going. 🧵
If you wanna keep those images for longer than an hour, you'll need to store them yourself. @cloudfare workers are a perfect fit for that.
I wrote a guide to get you going. 🧵