Prev:Dev advocate at GitHub. Graduated from Resilient Coders.
In my free time, I like to annoy my husband and binge watch reality tv shows.
I'm a new mom.
#BlackSky
This post explores how @blackgirlbytes.bsky.social used Agent Skills to encode design preferences and where human judgment still mattered most.
https://block.github.io/goose/blog/2026/01/04/how-i-taught-my-agent-my-design-taste
This post explores how @blackgirlbytes.bsky.social used Agent Skills to encode design preferences and where human judgment still mattered most.
https://block.github.io/goose/blog/2026/01/04/how-i-taught-my-agent-my-design-taste
Not really.
But I did realize I developed my taste through pattern recognition. Looking at what’s popular, what’s overdone, and what resonates.
So I taught goose my preferences, so I could have output I actually liked
block.github.io/goose/blog/2...
Not really.
But I did realize I developed my taste through pattern recognition. Looking at what’s popular, what’s overdone, and what resonates.
So I taught goose my preferences, so I could have output I actually liked
block.github.io/goose/blog/2...
and check out my repo to see the transcripts I had with my agent
and check out my repo to see the transcripts I had with my agent
The entire process is automated from end to end using goose recipes, Agent skills, shell scripts, and github actions.
#genuary #genuary3
The entire process is automated from end to end using goose recipes, Agent skills, shell scripts, and github actions.
#genuary #genuary2
The entire process is automated from end to end using goose recipes, Agent skills, shell scripts, and github actions.
#genuary #genuary2
- day 1
- day 2
- day 3
later on today..
- day 4
from me.
And then in the future you will see one post per day.
The entire process is automated from end to end using goose recipes, Agent skills, shell scripts, and github actions.
- day 1
- day 2
- day 3
later on today..
- day 4
from me.
And then in the future you will see one post per day.
The entire process is automated from end to end using goose recipes, Agent skills, shell scripts, and github actions.
The entire process is automated from end to end using goose recipes, Agent skills, shell scripts, and github actions.
It’s time for a mindset shift.
You’re not behind in AI.
You’re right on time.
It’s time for a mindset shift.
You’re not behind in AI.
You’re right on time.
dev.to/blackgirlbyt...
dev.to/blackgirlbyt...
dev.to/blackgirlbyt...
dev.to/blackgirlbyt...
You'll make bad art, you'll have people not like your work, and you'll be rejected. Moving past it is half the job of being a creative
You'll make bad art, you'll have people not like your work, and you'll be rejected. Moving past it is half the job of being a creative
We put goose to work on its own backlog by embedding it into a GitHub Action that turns issues into draft PRs.
I wrote about how this improved my day to day as a maintainer.
dev.to/blockopensou...
We put goose to work on its own backlog by embedding it into a GitHub Action that turns issues into draft PRs.
I wrote about how this improved my day to day as a maintainer.
dev.to/blockopensou...
These are getting compared a lot lately, and they shouldn’t be.
We wrote this post to clear up the confusion, and how it works within goose 👇
https://block.github.io/goose/blog/2025/12/21/code-mode-doesnt-replace-mcp
These are getting compared a lot lately, and they shouldn’t be.
We wrote this post to clear up the confusion, and how it works within goose 👇
https://block.github.io/goose/blog/2025/12/21/code-mode-doesnt-replace-mcp
https://block.github.io/goose/blog/2025/12/15/code-mode-mcp
https://block.github.io/goose/blog/2025/12/15/code-mode-mcp
This means MCP servers can delegate repeated AI decisions back to the client, without embedding or running their own models.
Day 12 of Advent of AI shows it in action.
Day 11's solution for Advent of AI is live. It's not too late to join the fun.
Link in the 🧵
Day 11's solution for Advent of AI is live. It's not too late to join the fun.
Link in the 🧵