Matt Pocock: Agent Skills and Engineering Workflows
New Skills! v1.2 brings /wait-what, /writing-for-agents, and fixes /grill-me
Seven videos on grilling, prototyping, planning, review, and software quality
A reviewed seven-video route through Matt Pocock’s agent-skills and engineering-workflow uploads from May 25 to August 19, 2026. Six tutorials cover Grill Me interviews, anti-slop review, repository-aware implementation, throwaway prototypes, Wayfinder decision maps, agent-facing writing, deterministic setup, and collaborator questionnaires. In a conversation hosted by Pocock, Uncle Bob Martin argues for deterministic quality gates, specialized agents, and modular architecture. The demonstrations and productivity claims are practitioner evidence, not controlled comparisons.
In this collection Browse 7 summaries 2 of 7
Matt Pocock's version 1.2 skills update targets friction in agent work: answers that are hard to read, design interviews slowed by one question per turn, and setup steps that need human control. He explains the revised instructions and shows selected interfaces and example outputs.
Key Points Covered
- Improve routing and installation metadata: The release adds a documentation site, a suggested workflow, marketplace packaging, and sidecar metadata intended to preserve invocation settings across supported harnesses 00:00–02:55(opens in a new tab).
- Repair unclear agent output locally: Wait What is designed to restate one difficult response in simpler technical English while grounding vocabulary in the project's context file 02:55–03:58(opens in a new tab).
- Ask independent questions together: Grill Me previously asked one question per turn, making easy decisions slow. The revision groups questions into rounds, but waits when a question depends on an earlier answer. Pocock shows numbered questions with recommended answers, followed by another round based on the responses 03:58–06:44(opens in a new tab).
- Write instructions for their consumers: Writing for Agents broadens skill-writing guidance to files such as
AGENTS.mdandCLAUDE.md, with an emphasis on concise always-loaded context and extracting specialized procedures into skills 06:44–07:33(opens in a new tab). - Keep setup under human control: Wizard generates an interactive shell script that opens configuration pages, guides the user, and saves supplied values. Pocock says entered credentials go through the script rather than the model; he does not review the generated script's security in the walkthrough 07:33–08:37(opens in a new tab).
- Bring collaborators' answers back to the agent: To Questionnaire exports unresolved decisions into a document people can discuss outside the agent interface. Pocock's example uses Google Docs, then imports the agreed answers back into the conversation 08:37–10:41(opens in a new tab).
Full video: https://www.youtube.com/watch?v=gaDdrDdczO4(opens in a new tab)