Aidan Cunniffe retweetledi
Community has been cooking this week 🚀
New contributors added support for @kilocode and @Kimi_Moonshot as well as porting the AI-blame extension to the @jetbrains ecosystem.
English
Aidan Cunniffe
4.2K posts
Aidan Cunniffe
@aidandcunniffe
building git-ai. exited founder. marathon addict.
nyc + sf + internet Katılım Ekim 2012
1.1K Takip Edilen1K Takipçiler
excited to annouce the latest scores of OurAgent on OurAgentBench:
1. OurAgent
2. YourAgent
arxiv paper in bio
English
@lenny_enderle git-ai gives agents the git and the prompt history. it's pretty wild how much smarter our agents got. it's the only context engineering thing I've tried (and I tried a lot of them) that just worked. Love how the answer ended up being simple.
English
@aidandcunniffe git history as agent context is underused. the commit log knows more about why code exists than any doc. agents that can actually read and reason over it have a real edge
English
Seeing this trend a lot of this across teams that use Git AI: enterprises are investing in their own background agents. I was skeptical at first, but I get it now.
Building and optimizing the factory will be the work of platform teams for the next 5 years.
Praveen Neppalli@praveenTweets
Agentic software engineering adoption is on fire at @Uber. 1,800 code changes per week are now written entirely by Uber's internal background coding agent, and 95% of our engineers now use AI every month across all the tools we track. This is a real reset moment for engineering; it's one of the most exciting times to lead. This shift requires builders to be curious and hands-on. I’m incredibly lucky to be surrounded by a team that’s doing exactly that. The best part is that the strongest adoption isn’t being pushed top down from leadership announcements; it’s coming from engineers who are quietly experimenting, quietly shipping, and quietly pushing things forward. I love spending time with those engineers because there’s no substitute for being close to the work. Over the last few months, we leaned in hard, and the results have been phenomenal. The bigger shift: going agentic. 84% of AI users are now working with agent-style workflows, not just tab completion. Claude Code usage nearly doubled in 2 months (32% → 63%), while IDE-based tools have largely plateaued. Engineers are moving from accepting suggestions to delegating tasks. Even within traditional IDEs, ~70% of committed code is now AI-generated. Background agents are writing code autonomously. Our internal background coding agent went from <1% of all code changes to 8% in just a few months. There is zero human authoring. Engineers review and approve, but the code is written entirely by AI agents. The role of the engineer is shifting - from writing every line to architecting systems and reviewing AI-generated code. More to come from the @UberEng team in the coming days.
English
@benjitaylor as it was in the beginning and will be in the end.
English
Increasingly convinced that the tools that win will be the tools that meet you where you are, in whatever workflow you have
English
Cursor Cloud Agent debugging its own Cloud environment is pretty 🔥.
English
Claude Code is having trouble today. Haven't even submitted the prompt yet...
Creeped up from 4 -> 7 before I killed it
English
Semi-serious proposal if you're having Amazon-style problems with Coding Agents.
Aidan Cunniffe@aidandcunniffe
English
The one time I wanted the agent to write a fallback....it didn't
English
Click here to get access to our code review agent for yourself: cloudflare.com/en-ca/careers/…
Dillon Mulroy@dillon_mulroy
i'm convinced we have the best code review agent at cloudflare and it's not even close in terms of quality signal to noise
English
you know something special is about to happen when the fan's on @savarlamov's laptop has been at 100% all day.
No white noise machine needed in this office
English
that feeling when the team is cooking magic 🪄 and you get to taste it early.
English
When I was at Atlassian we thought the fact PRDs lived in our tools would help make our coding agents better, but it didn't really pan out because all the decisions that get made as you build the feature don't make it back there.
yenkel@yenkel
I mean, I've rarely seen such a detailed PRD that mixed product requirements with technical specification, was kept up to date over time, etc. you still have some spec, or you can iterate on the feature with the agent directly and the spec lives in prompt history and you can use @usegitai to keep around as future context but you get the gist about the high level, BDUF PRD being dead
English
Everyone talking about "hiring" Agents
The Agents: Hi I'm a [software developer|marketer|SDR|support rep]. I take no responsibility for anything I do, but for a limited time only I charge by keystroke.
Responsibility will be 10x harder to scale than intelligence.
English
great to see Amp marking up AI Blames with its new Git AI stAMP.
Git AI@usegitai
Looks like @AmpCode was here 🚀. The Git AI <> Amp integration is now live. Track all the code written by Amp — along with the prompt that generated them.
English
@Suhail Git AI solves the problem by letting the Agents talk to the agent who wrote any line of code and ask what’s up.
A lot better than markdown you have to keep up to date.
English
It feels like someone should make a post-git-hook where it asks the AI model to look at the diff of what you changed for a merged PR and update the repo’s various readmes and other documentation to make it easier for an LLM to be able to write code and reference things faster rather than reading every single line of source code that might be relevant constantly. The agents need their own docs.
English
What's the best way to give top oss contributors Codex or Claude Code credits?
I'm not too worried about them using the credits for other projects (being very selective). If they do it's accidental or inconsequential.
English