The blog post discusses advanced techniques for manipulating git repositories, specifically focusing on how to send a precise fragment of a git repository to a server for processing and then return the results. It highlights the l...
About:
No specific summary about the author is available from the provided metadata.
Website:
Specializations:
Incoming Links:
Subscribe to RSS:
In the second part of his self-education series, David Crawshaw discusses the use of agents in programming with large language models (LLMs). He defines an agent as a simple construct that allows LLMs to execute commands and recei...
The blog post discusses a series of mini-outages experienced by sketch.dev due to issues arising from code written by a large language model (LLM). Initially, a deployment seemed stable, but CPU spikes and performance degradation ...
5.8How I program with LLMs
2025-07-15 • productivity programming generative ai go
David Crawshaw shares his personal experiences using generative models, particularly LLMs, in programming over the past year. He finds that LLMs significantly enhance his productivity, especially through autocomplete, search capab...
The author recounts their experience of watching a basketball game while simultaneously developing a new feature for Sketch, which involves integrating a browser capability using Chromium. They describe the process of gathering in...
The blog post discusses the communication strategies involved in task-oriented interactions with both humans and AI agents. It outlines four main approaches: Push (providing information upfront), Pull (waiting for questions), Resp...
The blog post discusses the debate among developers regarding the use of containers for agentic coding versus coding on local machines. While the team acknowledges the comfort of local setups with personalized configurations, they...
The blog post discusses the importance of using large language models (LLMs) to enhance the process of prompt engineering. The author shares insights on how to effectively collaborate with LLMs by providing clear instructions, amp...
The blog post discusses the development of merde.ai, a tool designed to automate the resolution of git merge conflicts using a large language model (LLM). The author explains the challenges faced in achieving effective conflict re...
The blog post discusses the implementation of a lightweight merge queue using GitHub Actions for repositories with fast tests and infrequent commits. It emphasizes the importance of maintaining a clean main branch with passing tes...
The blog post discusses seven tasks that the author, Philip Zeyliger, no longer performs manually due to the assistance of Sketch, a design tool. These tasks include rebasing and merging code, creating Grafana dashboards, executin...
The blog post discusses the author's experience using a coding agent, Sketch, to enhance productivity while debugging a coding issue related to zombie processes in a Go application. The author describes the process of investigatin...
The blog post discusses the author's experience with code reviewing in Sketch, highlighting the functionality of an editable diff view that allows users to make direct edits while reviewing code. The author shares a personal anecd...
The article discusses the concept of Agent Experience (AX), which is the user experience from the perspective of AI agents. It highlights the importance of designing systems that cater to how AI agents process information, emphasi...
The post discusses effective techniques for prompting AI agents in coding tasks, highlighting methods such as making code changes, using checklists, allowing for early failures, employing multiple agents concurrently, providing go...
The blog post discusses the development of an AI Programming Assistant called Sketch, highlighting its simplicity and effectiveness in utilizing a language model (LLM) with tool use. The author explains the core functionality of t...
The blog post discusses the author's experience using a coding agent, Sketch, to enhance productivity while programming. The author describes a situation where they encountered zombie processes in their code and initially struggle...
The author discusses the process of reviewing code in Sketch, sharing a trick he learned from John Reese at Google to edit the right-hand side of the diff view. He also mentions the ability to leave comments for the agent in the d...
The author discusses adding a browser feature to Sketch during a basketball game. They used the 'Farm Out Step 0' strategy to research and implement the feature, and faced challenges with installing Chromium and sending images. Th...
0How I program with Agents
2025-06-09 • programming machine learning api ai agents sql orm github
The text discusses the use of agents in programming, defining agents as a for loop containing an LLM call. It explains how agents can be used to navigate code bases, handle API use, and improve dependency management. The author al...
0Prompt Engineering and the Taste Gap
2025-06-02 • engineering ai
The author discusses the importance of prompt engineering and the role of LLMs in writing and editing prompts. They emphasize the need for clear instructions, context, and judgment in refining prompts, and the importance of human ...
The text discusses the concept of Agent Experience (AX) as a user experience from an AI agent's point of view, emphasizing the importance of designing for how LLMs process information and the components of AX such as system prompt...
The blog post discusses the similarities and differences between task-oriented communication with humans and AI agents. It introduces a conceptual framework for these interactions, outlining three options: Push, Pull, and Respond....
0How I program with LLMs
2025-05-22 •
...