←
Cindy Sridharan11,832 posts
Cindy Sridharan avatar

Cindy Sridharan

@copyconstruct
◷ Joined July 2010▧ 1,484 Photos and videos
Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·5m

I still read every line of code I merge to main, btw. I believe that the benefits of reading (skimming) code far outweigh whatever productivity“gains” I might get by not doing so. Reading code also doesn’t *that* long, IME, but ymmv.

I think you have to really commit to not reading the code and be willing to spend the tokens. Whenever I try this, even a quick peek has me steering the agent toward designs I prefer and then I’m back to reviewing code line by line. I haven’t tried @copyconstruct’s idea of putting favorite books/blogs/OSS repos in AGENTS.md yet so that might be what I’m missing.

Marc Brooker avatar
⟳ Cindy Sridharan retweeted
Marc Brooker@MarcJBrooker·2h

New blog post, with an ice-cold take on some of the last week's systems news: brooker.co.za/blog/2026/11/1… The throughput of real-world systems can vary (a lot!) based on subtle properties of the workload they're handling. Load tests that don't match the workload are useless.

Igor Kupczyński avatar
⟳ Cindy Sridharan retweeted
Igor Kupczyński@igor_kupczynski·3h

great post, preach @copyconstruct Code is going to be written, read and debugged almost exclusively by agents going forward. Even if you, like me, find it hard to let go of old shibboleths like reading and understanding every line of code, the depth of understanding you glean after a perfunctory scan of the code is materially different than the understanding you had when you were manually writing code. copyconstruct.medium.com/how…

Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·Nov 17

Really, it just comes down to “be empirical when drawing conclusions, hew as closely as possible to facts, and have the right facts in the first place. Dig deeper when you don’t have conclusive evidence, settle for no less” This applies to debugging, perf analysis, design etc.

Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·Nov 17

I’ve been doing some performance testing lately, and without proper guardrails, agents can draw wildly wrong or misleading conclusions when reading CPU and memory profiles. This framework I originally proposed for agentic *debugging* works well for agentic perf analysis as well.

How to effectively debug with agents when you no longer read or understand the code In this post, I propose a debugging framework for agents that doesn’t rely on human knowledge of code to get to the right “root cause” The TL;DR are the screenshots. Link to post in the reply.

Post attachmentPost attachmentPost attachmentPost attachment
Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·Nov 6

CI as it currently exists (build all the code to run the full test suite) seems like a waste of compute, which can be better assigned to agents to analyze and test select parts of code but way more rigorously. My pre-merge tests don’t do full builds; just targeted micro-tests.

How do you deal with the CI time to run all those tests, etc?

Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·Nov 6

Revisiting my tweets from 6m ago is funny, because I don’t run into any of these problems now at all. What was missing in my setup 6 months ago was the verification process I now use. 70% of my daily token spend goes into verifying patches, and only ~20-30% into creating them.

Might be anecdotal, but just this week multiple AI generated PRs with subtle bugs got merged that required several additional days and a lot of manual verification to fix. “Velocity” isn’t going up when you consider how much effort was spent fixing things post-merge, pre-deploy.

Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·Nov 5

This is 100% true. Your reliability in general should be *higher* with agent written code, as measured by error rates, SLO violations, client regressions etc. If you’re having more or equal incidents now than when you were manually writing code, that’s likely a “skill issue”.

Agreed. We have shockingly few production errors these days. E.g. panic from a missing nil-check happens once every couple of months, and usually involves legacy code humans partially wrote. At this point, if I write it by hand you are going to get more bugs.

Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·Nov 3

How to design good verification loops? Well, that’s most of the job now, isn’t it? Last week, I finally landed a refactor I started in Jan. It’s good I waited b/c the Jan version was buggy af. This is the verification process I used for refactors (have different ones for perf)

Post attachmentPost attachmentPost attachmentPost attachment

How does one ensure verification loops are actually correct? Rigorous needs to take cost into consideration as well.

Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·Nov 2

Getting agents to produce “high-quality, optimized outputs in a language or domain you’re personally unfamiliar with” is an emergent skill. It doesn’t happen magically, but requires: 1. A change in mindset 2. Setting up the best domain references and rigorous verification loops

I still can't believe people are rewriting software into Rust for performance reasons without learning any Rust. I'm optimizing a Go program right now and using Agents to do the work, but I have to steer them a lot or they get stuck in local minima, but I can only steer them because I have 10 years of doing performance work in Go. If I was optimizing a Rust program I'd have to just keep saying: "make it faster" over and over again until it got stuck somewhere dumb.

Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·Oct 31

Returning from international travel, followed by jet lag, physical exhaustion, remnants of a throat infection I picked up when traveling, dealing with a case of misplaced luggage across 3 carriers, catching up on work on a quasi-zombie state … Yeah, I’m glad its the weekend ❤️

Post attachmentPost attachmentPost attachmentPost attachment
Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·Oct 24

Didn’t Google just blog about how they solve this today? cloud.google.com/blog/topics… Secrets are kept entirely out of the agent sandbox. A security proxy injects credentials when needed. The “secure proxy” route to secure agent infra and secrets seems to be a popular pattern.

Is anyone solving how we give secrets/passwords to agents in a way that doesn't expose them to the upstream provider? It seems like codemode could be used to pass the tokens around and you'd only need to approve it for where the token is actually consumed

Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·Oct 23

The reason I still review every line of code despite encoding all the software engineering best practices in my agentic workflow, is because sometimes it’s the *exceptions* to these generalized rules that work best. The agent isn’t discerning enough to make these calls (as yet)

I've been reviewing a lot of code written by Opus 5.5, and it still seems to make a lot of bad software engineering decisions. The kind that will accumulate/calcify into unmaintainable software over time. IMO even the agents will struggle to maintain vibecoded software.

Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·Oct 23

At Paris CDG, Terminal 2E, security check: Gate K: no need to take your electronics out Gate M: take all your electronics out and put it in a tray Me: I wasn’t asked to do it in Gate K Security: Gate K is Gate K, Gate M is Gate M Never seen this in any other airport. 😕

Rohit Jnagal avatar
⟳ Cindy Sridharan retweeted
Rohit Jnagal@jnagal·Oct 22

I came back to Google to work on AI infra. Instead we got pulled in to solving the problem of exploding use of agents across the company. Announcing Agent Compute Engine: bringing planet scale agentic infra to Google services and now to Google Cloud. cloud.google.com/blog/topics…

Cindy Sridharan avatar
Cindy Sridharan@copyconstruct·Oct 22

I found this unpublished blog post from 6y ago (look at the hand drawn diagram lol). I’d totally forgotten about this, but the key ideas here (embracing non-determinism as a first class citizen via “error kernel” centric design etc) work super well as guidelines for my agents.

Post attachmentPost attachmentPost attachmentPost attachment