Ricardo (@undef_obj) profile picture

Working on being a good person first and everything else second. Principal Engineer @awscloud Currently working on Kiro

Issaquah, WA
Joined March 2011
Photo by Ricardo (@undef_obj)Photo by Ricardo (@undef_obj)Photo by Ricardo (@undef_obj)Photo by Ricardo (@undef_obj)Photo by Ricardo (@undef_obj)Photo by Ricardo (@undef_obj)Photo by Ricardo (@undef_obj)Photo by Ricardo (@undef_obj)Photo by Ricardo (@undef_obj)
Ricardo retweeted
Tired of your agent asking for permission for everything? Approve for me is now live in Kiro IDE, CLI, and Web. When an action isn't covered by your rules, an LLM review steps in. Kiro approves routine actions, blocks what it flags as unsafe, and asks you when it isn't sure, without giving up the safeguard. Full breakdown 👉spr.ly/6018ByExGr
14
9
3
65
6,391
Ricardo retweeted
This is how I work now. I've been running my life (work and at home) in Kiro Crew. One of the biggest reasons I have used Crew so heavily was the ability to run multiple sessions in parallel. Jncreasingly I had started using what I called "infinite" sessions. Long lasting sessions that covered one area that I kept coming back to. Over the past few weeks these infinite sessions have been replaced by Kiro bots, which are now available to all Kiro Crew users kiro.dev/blog/crew-bots
3
11
1
40
4,187
Today we're adding one of the most requested capabilities to Kiro Crew: Kiro bots, long-running and proactive agents for distinct roles. When we open-sourced Kiro Crew, builders quickly started running dozens of agents in parallel. This also introduced a new challenge we heard from builders and experienced within our own teams. Devs were spending more time shepherding context between agents and checking which sessions finished or which needed further input. We built Kiro bots to address this; a bot is a new way to structure and drive agentic work. Now, builders can give a bot one responsibility, like build health, and it works proactively to meet your goals, sharing results in one ongoing chat while driving parallel sessions behind the scenes. Decisions, tradeoffs, and context persist in Notes you can inspect and edit at any time. Each bot gets only the tools its role needs, runs and monitors the sessions behind it, and checks with you when a decision matters while keeping other work moving. Over time, your crew of bots gets better at each part of your workflow, taking the self-improving nature of Kiro Crew to a new level by letting you delegate entire processes. Thank you to the builders and contributors whose feedback shaped this release. Set up your first bot: kiro.dev/blog/crew-bots/
3
13
1
79
4,988
Ricardo retweeted
Kiro is now available as a first class provider in T3 Code!
23
15
2
557
23,706
Ricardo retweeted
Kiro CLI 3.0 is out. Close your laptop while a cloud session keeps working. Review the spec before Kiro writes code. Hand bigger tasks to workflows. With Approve for me, let Kiro handle prompt approvals for safe actions. Full breakdown → spr.ly/6011BHLuVX
18
24
4
121
9,948
Ricardo retweeted
We asked @kirodotdev to prove things about its own harness in Lean. Across 26 functions, it proved 70 properties and found 18 that broke, each reproduced on the actual code. One proof: a sub-agent can never end up with more permission than its parent. kiro.dev/blog/harness-proper…
2
5
2
25
2,534
Ricardo retweeted
Kiro lives up to the hype. I might replace @cursor_ai with @kirodotdev as my IDE if they can fix a few things. ✓ Specialized planning tools: plan, design, spec, todo. ✓ Really great at executing and overseeing tasks. ✓ Groks your codebase well & implements quickly! 🆇 Need to reign in the planning phase. It wants to be (overly) thorough. 🆇 Queued tasks always failed. 🆇 Separates diffs and code files / tends to feel more cluttered than cursor. ### Timestamps 00:00:00 - Introduction: First look at Kiro, an Amazon Cursor competitor. 00:00:24 - First Impressions: Exploring the Kiro UI (Specs, Agent, Hooks, Steering). 00:01:17 - First Prompt: Asking Kiro to analyze the project and suggest a feature. 00:02:34 - Kiro's Suggestion: AI-powered file analysis and processing. 00:03:31 - Narrowing Scope: Deciding to implement content extraction from images. 00:04:26 - Starting a Spec Session: Kiro initiates a dedicated planning and specification process. 00:05:06 - Generating Requirements: Kiro creates a requirements document with user stories. 00:06:44 - Design Phase: Moving from requirements to technical design. 00:07:17 - Generating a Design Doc: Kiro outlines the architecture, schema changes, and API operations. 00:08:04 - Wasp Background Jobs: Discussing how Kiro's plan might integrate with Wasp's job queue feature. 00:10:33 - Simplifying the Plan: Telling Kiro to create a simpler, iterative implementation plan. 00:12:50 - Implementation Phase: Kiro breaks the work down into a detailed task list. 00:14:53 - Task 1: Update Database Schema: Kiro modifies `schema.prisma` and prepares for migration. 00:15:33 - Running CLI Commands: Manually running the database migration command. 00:18:03 - Task 2: Create Content Extraction Utility: Kiro starts creating the utility function to call the OpenAI Vision API. 00:18:19 - Kiro vs. Cursor: Initial thoughts on how Kiro's workflow compares to Cursor. 00:19:59 - Handling Errors: Kiro runs into compilation errors and attempts to fix them. 00:22:36 - Task 3: Create Server Operation: Kiro works on the backend operation for content extraction. 00:26:05 - AWS Issues: Skipping ahead after resolving AWS permissions issues manually. 00:26:27 - Live Demo: Demonstrating the completed feature by uploading an image and extracting text. 00:27:41 - Final Review: Overall impressions of Kiro, its strengths (spec/planning), weaknesses (bugs), and potential.
2
6
3
18
2,357
Ricardo retweeted
En mi empresa los nuevos proyectos se están haciendo con spec-driven development usando herramientas como Kiro. Donde antes te hacían falta 5 juniors ahora con un senior y la IA vas sobrado. Hoy más que nunca es obligatorio invertir. No quieras depender de un salario.
10
11
1
231
20,507
OPINION: I’m voting yes on I-645 to repeal Washington’s so-called millionaires’ tax, and I hope you will too. I have plenty of reasons, but one line in the law deserves a closer look: it caps charitable deductions at $100,000. mynorthwest.com/kiro-opinion…
7
32
1
138
3,830
Washington, I know this ballot is confusing as hell. And that is on purpose. Politicians in our state are trying to confuse you becuase they don't believe in democracy. The only reason we get to vote on this is that enough brave people stood up and demanded it. Otherwise, the governor would have signed this into law like the King of England, treating us like serfs with no rights and no power. Well, I won't stand for being tricked, lied to, and treated like a child. Just to be crystal clear, if you vote "Yes," it will repeal the illegal income tax. It will not reduce funding to schools or human services because the tax doesn't go into effect until 2029. So we aren't taking anything away. Starbucks left because of this. Microsoft has threatened to move jobs out of state. Amazon now employs nearly 15,000 fewer people in Seattle than at its peak. Seattle is second to last in the number of VC deals that closed in Q3. Holding a small number of companies and rich people hostage, when they have options, is a foolish strategy that is already backfiring spectacularly. The solution? Reduce taxes on businesses and income. Why even tax people under 25 at all? Give families with children a HUGE tax credit. Make this state an incredible place to do business and have a family. Encourage entrepreneurs to come here. Get startups, agencies, and smart, motivated people to build here. We can increase overall state revenue if more people make stuff, employ people, and create more value for everyone here. Not by chasing them out of the state and creating a hostile environment for business. Remember this: governments and taxes don't make jobs. Entrepreneurs do.
55
150
13
762
25,324
“When opponents put an initiative on the November ballot to repeal the tax and prevent lawmakers from passing similar income taxes in the future, the Democratic state Attorney General added an ominous description to discourage voters from supporting it.” wsj.com/opinion/washington-m…
1
11
1
52
2,365
Democratic state Senate candidates in Washington drop the pretense that the millionaire tax will apply only to the wealthy. on.wsj.com/3Wefoyu
4
13
2
49
9,435
Many developers run @kirodotdev's agent directly on their local machine so it can see the same filesystem, toolchain, and localhost services they do. That's what makes it effective for building and testing real projects. Today, developers manage that risk by reviewing every command before approving it. As agents take on longer tasks and developers run more sessions in parallel, that attention becomes the bottleneck. Local sandboxing changes the trust model: even after a command is approved, the operating system still limits what it can read and write on your machine, and which network destinations it can reach. The agent still runs natively on your machine, but the commands it executes are restricted by the kernel. You can turn sandboxing on or off mid-session without restarting. Available now on macOS, Linux, and Windows: kiro.dev/blog/local-sandboxi…
3
8
18
1,912
Ricardo retweeted
A man cannot prompt the same agent twice, for it is has compacted and is not the same agent, and he is not the same man
92
912
63
9,114
225,281
"If God had wanted man to play soccer, He wouldn't have given us arms." - Mike Ditka RIP to a legend
84
1,375
90
15,454
421,942
Unfortunately a lot of people are not very good at clearly describing: 1. What they want. 2. How to verify that it is right. Or as it used to be called: requirements.
I am surprised that people are surprised this is how I prompt Claude. Talk to Claude the way you would a coworker. There's no secret to prompting. There's no need to be overly scaffolded or prescriptive for most tasks -- give Claude a goal, and it will figure it out. Back in the Sonnet 3.5 days, your prompt mattered a lot. Nowadays, it's much more important to communicate to the model: 1. What you want it to do 2. How much effort you want it to spend 3. How it should verify that it did the right thing
11
18
4
270
9,238
Ricardo retweeted
Kiro 💜 Manufacturing A 113-year-old, third-generation family-owned manufacturer with 35 developers compressed years of planned software development into weeks, rewriting what a small engineering team can build. 8 production-ready outcomes were delivered in 2 days. Read the full story: spr.ly/6015BG221J
6
6
1
50
6,254
Ricardo retweeted
I’ve been using Kiro to build an apps, and two of them have already made it to production. Everyone is talking about Codex and Claude Code. But I think Kiro deserves a seat at the table. I wrote about why 👇
5
2
8
412
Ricardo retweeted
deleting the product it's funny how you can sometimes tell if a team culture is dysfunctional just by looking at how their app is laid out. behind every feature, or tab in a sidebar likely lies a PM or someone responsible for an OKR. the problem is when individual performance is measured by these metrics. it becomes the singular focus of the responsible person to make those numbers look good, since it determines whether they get rewarded or axed, even if it comes at the detriment of the overall quality of the product. this is why you mostly see features get added, but rarely removed. if you own a feature, then it being removed feels like an aura loss. this is even worse in the agentic era when the cost to add features has become so low. you can simply will a new feature into existence overnight and add another tab. but users will notice that you ship your org chart. teams that make people play silly games under the guise of performance management rarely build great products. to build a great product you need to first fix your culture
157
199
57
3,164
175,958
Strands Decider 2B, a small, open-source decision model, perfect for experimentation but capable enough for real work: strandsagents.com/blog/intro… We've open sourced the whole recipe, with all the training data, code, and log of what we learned along the way: github.com/strands-labs/stra…
12
37
3
217
17,975