Code as Agent Harness SKILL usage and obvious results, summarized in this article:
devinternals.substack.com/p/the-day-my-c…
You can also experiment the SKILL in daily development tasks:
github.com/ShahriyarR/cod…
devinternals.substack.com/p/the-day-my-c…
You can also experiment the SKILL in daily development tasks:
github.com/ShahriyarR/cod…
0
0
4
164
Auto-evolve the SKILL?
In this article, I have tried to share the process of using Karpathy's autoresearch to self-improve the "Code as Agent Harness" SKILL.
devinternals.substack.com/p/autoresearch…
Activate the SKILL and give it a try to develop some feature or to fix some bug:
github.com/ShahriyarR/cod…
In this article, I have tried to share the process of using Karpathy's autoresearch to self-improve the "Code as Agent Harness" SKILL.
devinternals.substack.com/p/autoresearch…
Activate the SKILL and give it a try to develop some feature or to fix some bug:
github.com/ShahriyarR/cod…
0
0
0
108
SKILL.md for the presented idea in the paper:
github.com/ShahriyarR/cod…
Also tested it with simple feature development:
devinternals.substack.com/p/code-as-agen…
github.com/ShahriyarR/cod…
Also tested it with simple feature development:
devinternals.substack.com/p/code-as-agen…
// Code as Agent Harness //
100+ page report on all things related to agent harnesses.
(bookmark it)
In particular, the survey summarizes methods and applications of code as agent harness.
This paper makes a strong case that code-as-harness might be the key to moving us ce harness engineering.
Is code all you need? Maybe.
Regardless, the paper argues that future systems must have the following four properties: executable, inspectable, stateful, and governed.
Paper: arxiv.org/abs/2605.18747
Learn to build effective AI agents in our academy: academy.dair.ai
0
0
1
149
Simplicity Analysis of the Typer project based on @johnmaeda's Laws of Simplicity:
devinternals.substack.com/p/assessing-ty…
devinternals.substack.com/p/assessing-ty…
1
0
0
75
Inspired by @LangChain DeepAgents CLI, created scaffold agent-tui: github.com/ShahriyarR/age…
Most important changes:
* Proper Python Layered Architecture with events.
* AgentAdapter: Translates events from the agent into actions the TUI can perform
* DeepAgentsAdapter for DeepAgents
Most important changes:
* Proper Python Layered Architecture with events.
* AgentAdapter: Translates events from the agent into actions the TUI can perform
* DeepAgentsAdapter for DeepAgents
1
0
1
94
@sydneyrunkle is it something that the community can be interested in? :)
There is also an interactive course as a teaching material:
shahriyarr.github.io/agent-tui/agen…
There is also an interactive course as a teaching material:
shahriyarr.github.io/agent-tui/agen…
0
0
0
53
Hosting Zero to Agent in Wildau on 28th of April: global build week.
ship a real agent with @v0 + @vercel. free, $6k+ prizes.
luma.com/yjvfhm8f
#ZerotoAgent
ship a real agent with @v0 + @vercel. free, $6k+ prizes.
luma.com/yjvfhm8f
#ZerotoAgent
0
1
4
126
I just registered for an .agent domain and joined the .agent community! get yours now and help shape the future of autonomous agents agentcommunity.org/join#7T8PCZES @agentcommunity_
0
0
3
76
Currently using Superpowers with @MiniMax_AI M2.7 + @opencode Go - I feel great improvement on the results of the "brainstorming" for Superpowers. So definitely OpenCode Go models(Minimax, GLM, Kimi) is big go go for nearly all tasks.
1
0
3
141
This is my first Substack article - initial thoughts about the importance of keeping things simple in Agentic era:
devinternals.substack.com/p/the-best-ai-…
> as and independent research - I want to experiment with "Laws of Simplicity" by @johnmaeda and its relation with AI Agentic.
devinternals.substack.com/p/the-best-ai-…
> as and independent research - I want to experiment with "Laws of Simplicity" by @johnmaeda and its relation with AI Agentic.
1
0
1
68
First obvious experiment is to have a specialized SKILL and benchmark framework:
github.com/AI-Provenance/…
github.com/AI-Provenance/…
0
0
0
44
I was there xddd
buying pizza slices for a meetup and call it a day?
we went the extra mile — booked stockholm's most luxurious spa for a weekend & asked ourselves...
how fast can a single person build & monetize something with @Lovable?
here's what happened.
polar.sh/blog/prompt-a-…
0
0
2
106
Shako retweeted
browsirai. MCP server that connects AI agents to your running Chrome over CDP.
not a headless browser. not a Playwright wrapper. your actual Chrome process, with all your cookies, sessions, and extensions intact.
the agent never sees a login screen. you're already authenticated d:
npx browsirai install
not a headless browser. not a Playwright wrapper. your actual Chrome process, with all your cookies, sessions, and extensions intact.
the agent never sees a login screen. you're already authenticated d:
npx browsirai install
16
23
135
32,554
Shako retweeted
two things:
1. you should apply for Sentry for Startups (check eligibility: sentry.io/for/startups/a…)
2. Seer is really cool (see proof below)
1. you should apply for Sentry for Startups (check eligibility: sentry.io/for/startups/a…)
2. Seer is really cool (see proof below)
Got accepted for the Sentry Startup Program, and after experimenting with @sentry Seer, now cooking something interesting 🙂
The Sentry Integration, which can assist Seer with more in-depth root cause analysis 🚀
@Codydearkland sweet addition to you Seer videos xd
1
1
10
1,193
Got accepted for the Sentry Startup Program, and after experimenting with @sentry Seer, now cooking something interesting 🙂
The Sentry Integration, which can assist Seer with more in-depth root cause analysis 🚀
@Codydearkland sweet addition to you Seer videos xd
The Sentry Integration, which can assist Seer with more in-depth root cause analysis 🚀
@Codydearkland sweet addition to you Seer videos xd
1
1
6
1,447
Time to build some serious infrastructure 🎆
We got accepted to the @nvidia Inception Program.
We are building Shared Agentic AI Memory Management.
We have our first product, DevMemory CLI, currently in beta, looking for beta testers.
github.com/AI-Provenance/…
We got accepted to the @nvidia Inception Program.
We are building Shared Agentic AI Memory Management.
We have our first product, DevMemory CLI, currently in beta, looking for beta testers.
github.com/AI-Provenance/…
0
0
5
120