Matt Pocock
PulseAugur coverage of Matt Pocock — every cluster mentioning Matt Pocock across labs, papers, and developer communities, ranked by signal.
- authored /grill-me 90%
- authored Find Skills‐Based Volunteers With Catchafire's Virtual Matching Platform 90%
- developed Skillselion 90%
- instance of /grill-me 70%
- developed by /grill-me 70%
- affiliated with Vercel Labs 70%
- developed /grill-me 70%
- used by Skillselion 70%
- affiliated with /grill-me 50%
- competes with Vercel Labs 50%
10 day(s) with sentiment data
-
AI coding workflow v3 integrates skills, replaces prompts with automated checks
The author has updated their AI coding workflow to version 3, integrating all steps into a single repository called `agent-skills`. This new version replaces individual prompts with reusable "skills" and implements auto…
-
Open-source skills boost AI coding agents into disciplined engineers
Open-source skill collections are enhancing AI coding agents, transforming them into more disciplined engineers. Projects like Jesse Vincent's Superpowers methodology, Matt Pocock's composable engineering skills, and Ad…
-
Matt Pocock's Claude Code Habits Highlight Human Role in AI Development
Matt Pocock shared his habits when using Claude for coding, emphasizing the human element over the AI agent itself. Key takeaways include the importance of rigorous self-questioning before starting projects, learning to…
-
AI agents struggle to create code abstractions, study finds
A study explored whether AI agents can effectively refactor code by abstracting repeated logic, a task humans often struggle with. The experiment involved agents like Claude Code and Codex working on a TypeScript codeba…
-
Claude Code's "prototype" skills: Pocock vs. Kowalski explained
Two distinct "prototype" skills for Claude Code, developed by Matt Pocock and Emil Kowalski, share the same name but perform different functions. Pocock's skill is designed for rapid iteration on UI variants or logic te…
-
AI agents for code review gain traction with industry experts
Robert Cecil Martin and Matt Pocock are exploring the concept of giving code review its own AI agent, moving beyond simple code completion tools. This approach focuses on leveraging AI for more nuanced tasks like identi…
-
Model Context Protocol (MCP) standardizes AI tool integration
The Model Context Protocol (MCP) is a new standard designed to improve how AI models interact with external tools and data. Unlike a traditional API, MCP provides structured context to models, enabling them to better un…
-
Developer builds Zenn-to-dev.to crossposting pipeline with Claude Code
A developer has created a pipeline to automatically crosspost articles from Zenn to DEV Community using Claude Code. The process involves local translation via Claude Code and then using GitHub Actions to publish, updat…
-
GitHub Copilot adds local sandboxing for enhanced AI security
GitHub has made local sandboxing for its Copilot AI assistant generally available, enhancing security for users by restricting file, network, and credential access. This feature aims to prevent AI agents from accessing …
-
Claude Code skills fail to load due to poor description optimization
Many Claude Code skills are failing to load because developers are not optimizing their descriptions for Claude's semantic matching capabilities. Claude primarily uses the description field to determine if a skill shoul…
-
Anthropic turns Claude Code into a platform with new Mods system
Anthropic has introduced a "Mods" system for Claude Code, allowing developers to customize the AI coding tool using JavaScript or TypeScript. This feature enables modifications to the interface, interception of tool cal…
-
New CLI tool simplifies switching between Claude Code and Codex CLI
A developer has created a command-line interface tool called `session-handover` to streamline the process of switching between different AI coding assistants, specifically Claude Code and Codex CLI. The tool automatical…
-
Developer creates tools to improve AI idea refinement and Q&A
A developer has created two new tools, Ideator and Guide, designed to enhance the interaction with AI models like Claude. Ideator helps refine vague ideas by asking targeted questions that lead to different product outc…
-
Developer uses smaller AI models with strong guardrails for coding
A developer is exploring the use of smaller, faster AI models for software development by implementing strong guardrails and leveraging existing code patterns. This approach aims to reduce token usage and costs associat…
-
Matt Pocock's skills workflow offers structured alternative to 'vibe coding'
Matt Pocock has developed a "skills workflow" that he argues is more effective than "vibe coding." This approach breaks down complex tasks into smaller, manageable skills, offering a more structured and less overwhelmin…
-
Reddit users debate verbosity in LLM skill definitions
A Reddit user is questioning the effectiveness of adding explanatory prose to LLM skill definitions, specifically within GitHub's SKILL.md files. The user wonders if justifications like "they go stale fast" genuinely im…
-
Cursor IDE users debate "Superpowers" vs "Agent Skills" AI features
This Reddit discussion centers on the capabilities of AI tools, specifically comparing "Superpowers" and "Agent Skills" in the context of the Cursor IDE. Users are debating which of these features offers superior functi…
-
Matt Pocock unveils /wayfinder AI skill for complex project planning
Matt Pocock has developed a new AI skill called /wayfinder, designed to assist users and AI agents in navigating complex project planning, especially when the final outcome is uncertain. This skill aims to alleviate the…
-
Matt Pocock launches AI Skills to boost developer tool effectiveness
Matt Pocock has developed a suite of AI "skills" called "AI Skills for Real Engineers" designed to enhance the effectiveness of AI tooling for software developers. These skills integrate with AI harnesses like Claude Co…
-
Claude Code adoption shifts to author-based installs, data shows
The Claude Code ecosystem is seeing a shift in how developers adopt new tools, with a growing trend of searching for and installing skills based on author names rather than individual skill functionalities. This is evid…