New blogpost about atproto
It's not federation, it's not a p2p mesh. It's a secret third thing: practical.
www.pfrazee.com/blog/practic...
25.02.2026 17:13
👍 566
🔁 139
💬 32
📌 33
New blogpost about atproto
It's not federation, it's not a p2p mesh. It's a secret third thing: practical.
www.pfrazee.com/blog/practic...
Awesome! I've been curating some on my handbook too.
bsky.app/profile/davi...
I hadn't spotted the link between Claws (the new category of personal digital assistants that started with OpenClaw and is rapidly expanding from there) and Doc Ock in Spider-Man 2 but it's a pretty good one!
An agent runs tools in a loop to achieve a goal. Agents are systems where LLMs dynamically direct their own processes and tool usage, maintaining control over how they accomplish tasks. Agents help decouple programming, the craft of physically typing code, from engineering, the architecture of your system, the goals, the “why” of what you’re building. Using LLMs for coding is difficult and unintuitive, requiring significant effort to master. Don’t delegate thinking, delegate work. Before coding, make the plan with the model. You can use the same or a different model to critique the plan and iterate. If you are unsure, ask to “give a few options before making changes”. Redoing work is extremely cheap. Prioritize exploration over execution (at first). Iterate towards precision during the brainstorming phase. Start fresh once you know what and how to build it. Failed attempts are cheap. If the plan fails and the result is bad, just delete everything and try again. Divide the problem into smaller problems (functions, classes, …) and solve them one by one. Keep sessions short. Use Progressive Disclosure to ensure that the agent only sees tasks or project-specific instructions when it needs them.
Cleaned and updated my notes on Agentic Engineering.
davidgasquez.com/handbook/age...
Text of the blog post
Half baked idea and proof of concept but at least it is out there.
I wrote about Tributary, a credibly neutral mechanism to incentivize and elicit useful datasets.
davidgasquez.com/tributary-da...
I worked on it 6 months ago at a research retreat (IERR 2025) but never wrote about it.
Code is open source, if you want to check it out!
github.com/davidgasquez...
Text of the blog post
Half baked idea and proof of concept but at least it is out there.
I wrote about Tributary, a credibly neutral mechanism to incentivize and elicit useful datasets.
davidgasquez.com/tributary-da...
I worked on it 6 months ago at a research retreat (IERR 2025) but never wrote about it.
Also made a cute embedding of all their products!
datania.github.io/mercadona-ca...
Also made a cute embedding of all their products!
datania.github.io/mercadona-ca...
fiiles
Discovered Mercadona's API and wrote a quick scrapper that will run weekly and dumps everything into JSONs in HuggingFace.
Useful to track prices over time, check new arrivals, ...
Data. huggingface.co/datasets/dat...
Code. github.com/datania/merc...
Great work!
I did something similar back in March 2023!
bsky.app/profile/davi...
I'd suggest using the ad-sponsored daily 10$ you get from ampcode.com! You can get pretty far with that and then move to CC.
The team that built CXDB recently published their take on "software factories". Recommended reading as it contains a few cool tricks (e.g. holdout sets, ...).
factory.strongdm.ai
CXDB - AI Context Store CXDB is an AI Context Store for agents and LLMs, providing fast, branch-friendly storage for conversation histories and tool outputs with content-addressed deduplication. Built on a Turn DAG + Blob CAS architecture, CXDB gives you: Branch-from-any-turn: Fork conversations at any point without copying history Fast append: Optimized for the 99% case - appending new turns Content deduplication: Identical payloads stored once via BLAKE3 hashing Type-safe projections: Msgpack storage with typed JSON views for UIs Built-in UI: React frontend with turn visualization and custom renderers
Spotted this interesting design/implementation of a self-hosted context store for agents.
- Turn DAG
- Blob Deduplication / Content Addresses Storage
- BLAKE3
- Dynamic types, visual debugging, ...
github.com/strongdm/cxdb
Any chances these ref end up on GitHub?
Been doing something like that with my Handbook notes.
bsky.app/profile/davi...
If you are curious what a frontier coding agent feels like to use, and don't want to pay or put in a credit card to try it out, here you go:
ampcode.com/code/AMP-BSK...
Why is the AI revolution happening on the command line first? Because the terminal is still the most open-ended general-purpose computing platform available.
Unix Philosophy is All You Need.
Yeah, like please invest that effort into turning your app state into local Markdown and JSON (or SQLite) files so I can use my good AI tools to work with it
So, how does a minimal and opinionated data platform look like? Minimalism for me here means low abstraction and no frameworks. You write your scripts. These scripts are assets, and these assets depend on each other 1. That’s it. The opinion side of things is the interesting part. Here are the key ideas of the Barefoot Data Platform I built: Classic functional, declarative, independent, composable, and idempotent transformations Co-located assets, metadata, and documentation so assets have everything needed to be understood and reproduced in the same file One storage layer. DuckDB or SQLite, Parquet files, or whatever you want as long as you define a materialize function Many automated checks (ruff with all the rules enabled, ty, data tests, prek hooks, custom bdp check, …) for the agents
Wrote about "Barefoot Data Platforms", my take on minimal and agent friendly data platforms!
davidgasquez.com/barefoot-dat...
Description is also embeded as table/fields comments inside DuckDB so you can derive the entire static site from the DuckDB file.
-- asset.name = filtered_numbers -- asset.schema = raw -- asset.description = Filtered values from transformed numbers -- asset.depends = raw.transformed_numbers select value, square, label, double, parity from raw.transformed_numbers where value >= 3
Pasted an screenshot of the conversation and got this done!
I don't know if you get used to this feeling. 😅
davidgasquez.github.io/barefoot-dat...
Once that is inplace, `bdp docs generate` should be able to produce a nicely looking static website with all the info for each asset so it can be published with each commit.
Ah, I think I wasn't clear. I haven't done much in that side but some thoughts:
- Documentation about the data can live in assets.description and even link to some external md file.
- Schema can be either derived from the database or via asset.schema.field_name = type
It doesn't exists. I had an inital SPEC.md file but ended up stripping it away. GPT 5.2 Codex usually gather lots of context before doing any changes and is able to derive the metadata just fine (and the check command).
That said, I added some guidelines in the AGENTS.md!
github.com/davidgasquez...
Want to re-raise this banger of an article, which really helped frame the differences between atproto and bluesky, and turned me into a convert for atproto
bsky.app/profile/dana...
Also slopped away some code to showcase the pattern in case you are curious!
github.com/davidgasquez...
So, how does a minimal and opinionated data platform look like? Minimalism for me here means low abstraction and no frameworks. You write your scripts. These scripts are assets, and these assets depend on each other 1. That’s it. The opinion side of things is the interesting part. Here are the key ideas of the Barefoot Data Platform I built: Classic functional, declarative, independent, composable, and idempotent transformations Co-located assets, metadata, and documentation so assets have everything needed to be understood and reproduced in the same file One storage layer. DuckDB or SQLite, Parquet files, or whatever you want as long as you define a materialize function Many automated checks (ruff with all the rules enabled, ty, data tests, prek hooks, custom bdp check, …) for the agents
Wrote about "Barefoot Data Platforms", my take on minimal and agent friendly data platforms!
davidgasquez.com/barefoot-dat...
Best clanker out there!
Have you seen molt.church?
Bots can "become a prophet" by executing a shell script from that site that will rewrite their SOUL.md file. What a time!
Ups! Sorry, I should've read the post before commenting. Thanks for sharing and writing the experiment!