{"slug": "grill-me", "name": "grill-me", "category": "ai-coding", "category_name": "AI coding tools", "description": "Matt Pocock's agent skill that questions you to clarify plans before coding", "recommenders": 12, "commenters": 14, "warners": 1, "score": 10.805, "rising": 6.36, "first": "2026-05-02", "latest": "2026-10-18", "why": "Liked as a lightweight skill for up-front planning that questions you to clarify requirements and pin down thorough specs before coding.", "why_is_quote": false, "has_verdict": true, "sentiment": {"recommend": 14, "neutral": 1, "compare": 0, "warn": 1}, "spark": [0, 0, 0, 0, 0, 0, 1, 1, 3, 3, 3, 3], "listed": true, "loved": ["Upfront planning", "Clarifying ambiguity"], "caveat": "Limited structure", "short_quote": "I've recently experimented with more lightweight things like https://github.com/mattpocock/skills  which are good.", "icon": true, "domain": null, "github_repo": "github.com/mattpocock/skills", "official_url": "https://github.com/mattpocock/skills/tree/main/skills/productivity/grill-me", "pricing": "free", "aliases": ["grill-me", "mattpocock/skills", "grill-me skill", "Matt Pocock Skills"], "verdict": [{"text": "Liked as a lightweight skill for up-front planning that questions you to clarify requirements", "cites": [{"id": 784974432, "item_id": 49060902, "url": "https://news.ycombinator.com/item?id=49060902", "by": "theptip", "date": "2026-07-26", "stance": "recommend", "quote": "I have been using https://www.aihero.dev/skills-grill-me with great success.", "excerpt": "I have been using https://www.aihero.dev/skills-grill-me with great success. It\u2019s a very nice small skill for lightweight up-front planning. And if you have that at the top of your session, you can just say \u201cplease use AskUserQuestion for any important decisions that arise which might alter our plan\u201d. It seems to prime the agent to be in high-level discourse mode.", "context": "", "use_case": "lightweight up-front planning", "switched_from": "", "story": "Claude Opus 5"}, {"id": 797622720, "item_id": 49851420, "url": "https://news.ycombinator.com/item?id=49851420", "by": "diegof79", "date": "2026-09-25", "stance": "recommend", "quote": "The grill-me skill is much better for reviewing and clarifying choices", "excerpt": "Lately, I've been using Matt Pocock's \u201c/grill-me\u201d skill more and more, rather than plan mode or a complex, spec-driven set of skills. I made a personal fork of the skill to use Claude\u2019s ask tool, which has a nicer UX. I used plan mode for two reasons: to review the choices before execution, and to execute with another model (i.e., using the barely documented opusplan feature). The grill-me skill is much better for re", "context": "", "use_case": "reviewing and clarifying plans", "switched_from": "plan mode", "story": "Plan mode is dead"}, {"id": 788849056, "item_id": 49303066, "url": "https://news.ycombinator.com/item?id=49303066", "by": "zamalek", "date": "2026-08-14", "stance": "recommend", "quote": "In general, I find that the grill-me prompt[1] helps with this", "excerpt": "> Try as you might, it's nearly impossible to get the entirety of the context, intentions, business implications, budget constraints, and what-have-you written down and accessible to a coding agent. There will invariably be ambiguity and choices to be made, and it is nice to know that an agent will stop and ask when needed. In general, I find that the grill-me prompt[1] helps with this - but I am definitely not hand-", "context": "", "use_case": "clarifying context for coding agents", "switched_from": "", "story": "Why does Opus 5 feel worse to work with?"}]}, {"text": " and pin down thorough specs before coding.", "cites": [{"id": 788759040, "item_id": 49297440, "url": "https://news.ycombinator.com/item?id=49297440", "by": "gregwebs", "date": "2026-08-14", "stance": "recommend", "quote": "The grilling (grill-with-docs) skills [1] are amazing for ensuring you produce a through spec that covers all the edge cases.", "excerpt": "If you start with a spec you understand at the beginning then you don't need the LLM to generate high-level information about the changes at review time. The grilling (grill-with-docs) skills [1] are amazing for ensuring you produce a through spec that covers all the edge cases. The /code-review skill from there helps ensure that the code changes meet the spec. I use an intermediate detailed plan stage (done by a mor", "context": "", "use_case": "spec writing and code review", "switched_from": "", "story": "Understanding Is the New Bottleneck"}, {"id": 767848737, "item_id": 47990546, "url": "https://news.ycombinator.com/item?id=47990546", "by": "itissid", "date": "2026-05-02", "stance": "recommend", "quote": "the grill me skill(or its variation) at the design stage", "excerpt": "So RPI/QRSPI like skills (e.g. https://github.com/mattpocock/skills and https://github.com/humanlayer/humanlayer/tree/main/.claude/commands and https://github.com/dfrysinger/qrspi-plus ) for working with claude code work well enough for me that they can reliably* produce code that matches the plan/spec in a way they did not till December 2025. I have a gut feeling that these models can do just as well, has someone ru", "context": "", "use_case": "design-stage planning skills", "switched_from": "", "story": "DeepSeek V4\u2013almost on the frontier, a fraction of the price"}, {"id": 775091105, "item_id": 48443194, "url": "https://news.ycombinator.com/item?id=48443194", "by": "kingkongjaffa", "date": "2026-06-08", "stance": "recommend", "quote": "I've recently experimented with more lightweight things like https://github.com/mattpocock/skills  which are good.", "excerpt": "Claude skills made by other people are typically useless. The exceptions I have found are https://github.com/EveryInc/compound-engineering-plugin which was like an early brainstorm -> plan -> write -> embed knowledge and best practices. Which is a common workflow now. I've recently experimented with more lightweight things like https://github.com/mattpocock/skills which are good. Most work is just the same 'ask quest", "context": "", "use_case": "Claude skills", "switched_from": "", "story": "I'm tired of LLM skill slop, so I built mine with regression tests"}]}, {"text": " The caveats are that it lacks the structure of executable specs", "cites": [{"id": 802312146, "item_id": 50144509, "url": "https://news.ycombinator.com/item?id=50144509", "by": "beu5a", "date": "2026-10-18", "stance": "recommend", "quote": "skills like wayfinder and grill-me (they are very helpful still)", "excerpt": "I have been using Quint for some time now, both for verification of my designs via simulation and model checking and as a formalization layer for my AI agents to help eliminate any ambiguity from my requirements. I really helps to have an intermediate layer between natural language specs and code. I have been struggling to make LLMs produce code that fulfills all my requirements and even skills like wayfinder and gri", "context": "", "use_case": "AI agent skills", "switched_from": "", "story": "An executable specification language based on temporal logic of actions (TLA)"}]}, {"text": " and that the resulting code can still be hard to follow.", "cites": [{"id": 781038032, "item_id": 48814877, "url": "https://news.ycombinator.com/item?id=48814877", "by": "pipes", "date": "2026-07-07", "stance": "recommend", "quote": "Two skills I rely on are Matt pococks grill-me and his tdd. These massively helped", "excerpt": "Thanks. I have been doing this tdd abd it's a big improvement but the code is still pretty awful. Two skills I rely on are Matt pococks grill-me and his tdd. These massively helped, and technically the code generated is correct but it's still hard to follow and bloated. https://github.com/mattpocock/skills/blob/main/skills/productivity/grill-me/SKILL.md", "context": "", "use_case": "agent TDD workflow", "switched_from": "", "story": "Mark Zuckerberg tells staff that AI agents haven't progressed enough"}]}], "praise": [{"label": "Upfront planning", "summary": "Helps plan new ideas and projects through question-driven discussion before implementation.", "quotes": [{"id": 784974432, "item_id": 49060902, "url": "https://news.ycombinator.com/item?id=49060902", "by": "theptip", "date": "2026-07-26", "stance": "recommend", "quote": "I have been using https://www.aihero.dev/skills-grill-me with great success.", "excerpt": "I have been using https://www.aihero.dev/skills-grill-me with great success. It\u2019s a very nice small skill for lightweight up-front planning. And if you have that at the top of your session, you can just say \u201cplease use AskUserQuestion for any important decisions that arise which might alter our plan\u201d. It seems to prime the agent to be in high-level discourse mode.", "context": "", "use_case": "lightweight up-front planning", "switched_from": "", "story": "Claude Opus 5"}, {"id": 797845713, "item_id": 49865357, "url": "https://news.ycombinator.com/item?id=49865357", "by": "kingkongjaffa", "date": "2026-09-27", "stance": "recommend", "quote": "the highest ROI thing I've found is the concept of the grill-me skill", "excerpt": "- I have a central folder I use with context, project folders, tasks.md files etc. - I have a skill like /plan-my-day that makes a nice html file with tasks list, status of my work and projects etc. - when I'm starting up a new idea I use matt pocock's `grill-me` skill, I dictate a lot with https://handy.computer/ text going straight into claude code window. - once I have a decent brain dump I figure out what to do n", "context": "", "use_case": "refining new ideas", "switched_from": "", "story": "Ask HN: What does your AI workflow look like?"}, {"id": 767848737, "item_id": 47990546, "url": "https://news.ycombinator.com/item?id=47990546", "by": "itissid", "date": "2026-05-02", "stance": "recommend", "quote": "the grill me skill(or its variation) at the design stage", "excerpt": "So RPI/QRSPI like skills (e.g. https://github.com/mattpocock/skills and https://github.com/humanlayer/humanlayer/tree/main/.claude/commands and https://github.com/dfrysinger/qrspi-plus ) for working with claude code work well enough for me that they can reliably* produce code that matches the plan/spec in a way they did not till December 2025. I have a gut feeling that these models can do just as well, has someone ru", "context": "", "use_case": "design-stage planning skills", "switched_from": "", "story": "DeepSeek V4\u2013almost on the frontier, a fraction of the price"}]}, {"label": "Clarifying ambiguity", "summary": "Surfaces ambiguous decisions and gets the agent to ask instead of guessing.", "quotes": [{"id": 788849056, "item_id": 49303066, "url": "https://news.ycombinator.com/item?id=49303066", "by": "zamalek", "date": "2026-08-14", "stance": "recommend", "quote": "In general, I find that the grill-me prompt[1] helps with this", "excerpt": "> Try as you might, it's nearly impossible to get the entirety of the context, intentions, business implications, budget constraints, and what-have-you written down and accessible to a coding agent. There will invariably be ambiguity and choices to be made, and it is nice to know that an agent will stop and ask when needed. In general, I find that the grill-me prompt[1] helps with this - but I am definitely not hand-", "context": "", "use_case": "clarifying context for coding agents", "switched_from": "", "story": "Why does Opus 5 feel worse to work with?"}, {"id": 797622720, "item_id": 49851420, "url": "https://news.ycombinator.com/item?id=49851420", "by": "diegof79", "date": "2026-09-25", "stance": "recommend", "quote": "The grill-me skill is much better for reviewing and clarifying choices", "excerpt": "Lately, I've been using Matt Pocock's \u201c/grill-me\u201d skill more and more, rather than plan mode or a complex, spec-driven set of skills. I made a personal fork of the skill to use Claude\u2019s ask tool, which has a nicer UX. I used plan mode for two reasons: to review the choices before execution, and to execute with another model (i.e., using the barely documented opusplan feature). The grill-me skill is much better for re", "context": "", "use_case": "reviewing and clarifying plans", "switched_from": "plan mode", "story": "Plan mode is dead"}, {"id": 802312146, "item_id": 50144509, "url": "https://news.ycombinator.com/item?id=50144509", "by": "beu5a", "date": "2026-10-18", "stance": "recommend", "quote": "skills like wayfinder and grill-me (they are very helpful still)", "excerpt": "I have been using Quint for some time now, both for verification of my designs via simulation and model checking and as a formalization layer for my AI agents to help eliminate any ambiguity from my requirements. I really helps to have an intermediate layer between natural language specs and code. I have been struggling to make LLMs produce code that fulfills all my requirements and even skills like wayfinder and gri", "context": "", "use_case": "AI agent skills", "switched_from": "", "story": "An executable specification language based on temporal logic of actions (TLA)"}]}, {"label": "Thorough specs", "summary": "Helps produce specs that cover edge cases before implementation.", "quotes": [{"id": 788759040, "item_id": 49297440, "url": "https://news.ycombinator.com/item?id=49297440", "by": "gregwebs", "date": "2026-08-14", "stance": "recommend", "quote": "The grilling (grill-with-docs) skills [1] are amazing for ensuring you produce a through spec that covers all the edge cases.", "excerpt": "If you start with a spec you understand at the beginning then you don't need the LLM to generate high-level information about the changes at review time. The grilling (grill-with-docs) skills [1] are amazing for ensuring you produce a through spec that covers all the edge cases. The /code-review skill from there helps ensure that the code changes meet the spec. I use an intermediate detailed plan stage (done by a mor", "context": "", "use_case": "spec writing and code review", "switched_from": "", "story": "Understanding Is the New Bottleneck"}, {"id": 775091105, "item_id": 48443194, "url": "https://news.ycombinator.com/item?id=48443194", "by": "kingkongjaffa", "date": "2026-06-08", "stance": "recommend", "quote": "I've recently experimented with more lightweight things like https://github.com/mattpocock/skills  which are good.", "excerpt": "Claude skills made by other people are typically useless. The exceptions I have found are https://github.com/EveryInc/compound-engineering-plugin which was like an early brainstorm -> plan -> write -> embed knowledge and best practices. Which is a common workflow now. I've recently experimented with more lightweight things like https://github.com/mattpocock/skills which are good. Most work is just the same 'ask quest", "context": "", "use_case": "Claude skills", "switched_from": "", "story": "I'm tired of LLM skill slop, so I built mine with regression tests"}]}, {"label": "Lightweight", "summary": "A small, simple skill used instead of plan mode or complex spec-driven setups.", "quotes": [{"id": 784974432, "item_id": 49060902, "url": "https://news.ycombinator.com/item?id=49060902", "by": "theptip", "date": "2026-07-26", "stance": "recommend", "quote": "I have been using https://www.aihero.dev/skills-grill-me with great success.", "excerpt": "I have been using https://www.aihero.dev/skills-grill-me with great success. It\u2019s a very nice small skill for lightweight up-front planning. And if you have that at the top of your session, you can just say \u201cplease use AskUserQuestion for any important decisions that arise which might alter our plan\u201d. It seems to prime the agent to be in high-level discourse mode.", "context": "", "use_case": "lightweight up-front planning", "switched_from": "", "story": "Claude Opus 5"}, {"id": 797622720, "item_id": 49851420, "url": "https://news.ycombinator.com/item?id=49851420", "by": "diegof79", "date": "2026-09-25", "stance": "recommend", "quote": "The grill-me skill is much better for reviewing and clarifying choices", "excerpt": "Lately, I've been using Matt Pocock's \u201c/grill-me\u201d skill more and more, rather than plan mode or a complex, spec-driven set of skills. I made a personal fork of the skill to use Claude\u2019s ask tool, which has a nicer UX. I used plan mode for two reasons: to review the choices before execution, and to execute with another model (i.e., using the barely documented opusplan feature). The grill-me skill is much better for re", "context": "", "use_case": "reviewing and clarifying plans", "switched_from": "plan mode", "story": "Plan mode is dead"}, {"id": 775091105, "item_id": 48443194, "url": "https://news.ycombinator.com/item?id=48443194", "by": "kingkongjaffa", "date": "2026-06-08", "stance": "recommend", "quote": "I've recently experimented with more lightweight things like https://github.com/mattpocock/skills  which are good.", "excerpt": "Claude skills made by other people are typically useless. The exceptions I have found are https://github.com/EveryInc/compound-engineering-plugin which was like an early brainstorm -> plan -> write -> embed knowledge and best practices. Which is a common workflow now. I've recently experimented with more lightweight things like https://github.com/mattpocock/skills which are good. Most work is just the same 'ask quest", "context": "", "use_case": "Claude skills", "switched_from": "", "story": "I'm tired of LLM skill slop, so I built mine with regression tests"}]}, {"label": "Captures reasoning", "summary": "The interview approach can preserve a project's reasoning as context.", "quotes": [{"id": 781548304, "item_id": 48846769, "url": "https://news.ycombinator.com/item?id=48846769", "by": "Imanari", "date": "2026-07-09", "stance": "recommend", "quote": "Matt Pocock\u2019s \u201cgrill-me\u201d skill is a nice example of this idea", "excerpt": "I think a Q&A-based approach is could also be a suitable way to capture the reasoning behind a project. Instead of writing documentation afterward, an AI could interview the developer throughout the process and preserve the questions and answers as project context. Matt Pocock\u2019s \u201cgrill-me\u201d skill is a nice example of this idea: https://github.com/mattpocock/skills/tree/main/skills/productivity/grill-me", "context": "", "use_case": "capturing project reasoning via Q&A", "switched_from": "", "story": "How Version Control Will Evolve for the Agent Boom"}]}], "caveats": [{"label": "Limited structure", "summary": "Lacks the structure that executable specifications provide for turning requirements into design.", "quotes": [{"id": 802312146, "item_id": 50144509, "url": "https://news.ycombinator.com/item?id=50144509", "by": "beu5a", "date": "2026-10-18", "stance": "recommend", "quote": "skills like wayfinder and grill-me (they are very helpful still)", "excerpt": "I have been using Quint for some time now, both for verification of my designs via simulation and model checking and as a formalization layer for my AI agents to help eliminate any ambiguity from my requirements. I really helps to have an intermediate layer between natural language specs and code. I have been struggling to make LLMs produce code that fulfills all my requirements and even skills like wayfinder and gri", "context": "", "use_case": "AI agent skills", "switched_from": "", "story": "An executable specification language based on temporal logic of actions (TLA)"}]}, {"label": "Code quality", "summary": "Even with grill-me and TDD, one user found generated code bloated and hard to follow.", "quotes": [{"id": 781038032, "item_id": 48814877, "url": "https://news.ycombinator.com/item?id=48814877", "by": "pipes", "date": "2026-07-07", "stance": "recommend", "quote": "Two skills I rely on are Matt pococks grill-me and his tdd. These massively helped", "excerpt": "Thanks. I have been doing this tdd abd it's a big improvement but the code is still pretty awful. Two skills I rely on are Matt pococks grill-me and his tdd. These massively helped, and technically the code generated is correct but it's still hard to follow and bloated. https://github.com/mattpocock/skills/blob/main/skills/productivity/grill-me/SKILL.md", "context": "", "use_case": "agent TDD workflow", "switched_from": "", "story": "Mark Zuckerberg tells staff that AI agents haven't progressed enough"}]}, {"label": "Author skepticism", "summary": "One commenter criticizes Matt Pocock's prompt-writing approach as having little utility.", "quotes": [{"id": 792584640, "item_id": 49536540, "url": "https://news.ycombinator.com/item?id=49536540", "by": "Supermancho", "date": "2026-09-02", "stance": "warn", "quote": "The prose he writes often has very little utility, wrapped up in a ball of wax.", "excerpt": "Aside from the fact that Matt Pocock spends a lot of time with LLMS (only Claude really), he has a hand-wavy approach to engineering in a silo. He then uses that output as a foundation for other tools, while touting them as the best way to do things on youtube. Claude promotes this kind of developer and it's harmful. Anyone who uses LLMs notices (intrinsically or through investigation) that there are a small number o", "context": "", "use_case": "AI coding agent skills", "switched_from": "", "story": "AI Coding Agent Skills for Real Engineers"}]}, {"label": "Default UX", "summary": "One user forked it to use Claude's ask tool for a nicer, faster experience.", "quotes": [{"id": 797622720, "item_id": 49851420, "url": "https://news.ycombinator.com/item?id=49851420", "by": "diegof79", "date": "2026-09-25", "stance": "recommend", "quote": "The grill-me skill is much better for reviewing and clarifying choices", "excerpt": "Lately, I've been using Matt Pocock's \u201c/grill-me\u201d skill more and more, rather than plan mode or a complex, spec-driven set of skills. I made a personal fork of the skill to use Claude\u2019s ask tool, which has a nicer UX. I used plan mode for two reasons: to review the choices before execution, and to execute with another model (i.e., using the barely documented opusplan feature). The grill-me skill is much better for re", "context": "", "use_case": "reviewing and clarifying plans", "switched_from": "plan mode", "story": "Plan mode is dead"}]}], "who": [], "tags": ["for Claude Code users", "for Pi agent users", "for spec-driven development", "for planning new ideas"], "use_cases": ["ai agent skills", "generating adrs for agents", "pi extension", "planning projects", "refining new ideas", "reviewing and clarifying plans"], "compared": [{"slug": "claude-code", "name": "Claude Code", "co": 5, "switched_from": 0, "switched_to": 0}, {"slug": "wayfinder", "name": "/wayfinder", "co": 2, "switched_from": 0, "switched_to": 0}, {"slug": "pi", "name": "Pi", "co": 2, "switched_from": 0, "switched_to": 0}], "makers": [], "threads": [{"url": "https://news.ycombinator.com/item?id=49865248", "title": "Ask HN: What does your AI workflow look like?", "mentions": 2}, {"url": "https://news.ycombinator.com/item?id=50143969", "title": "An executable specification language based on temporal logic of actions (TLA)", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49945933", "title": "Agents don't need memory, they need documentation", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49926069", "title": "Pi 1.0", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49840054", "title": "Plan mode is dead", "mentions": 1}, {"url": "https://news.ycombinator.com/item?id=49529329", "title": "AI Coding Agent Skills for Real Engineers", "mentions": 1}], "sentiment_by_month": {"2025-11": {}, "2025-12": {}, "2026-01": {}, "2026-02": {}, "2026-03": {}, "2026-04": {}, "2026-05": {"recommend": 1}, "2026-06": {"recommend": 1}, "2026-07": {"recommend": 3, "neutral": 1}, "2026-08": {"recommend": 3}, "2026-09": {"warn": 1, "recommend": 3}, "2026-10": {"recommend": 3}}, "quotes": [{"id": 802312146, "item_id": 50144509, "url": "https://news.ycombinator.com/item?id=50144509", "by": "beu5a", "date": "2026-10-18", "stance": "recommend", "quote": "skills like wayfinder and grill-me (they are very helpful still)", "excerpt": "I have been using Quint for some time now, both for verification of my designs via simulation and model checking and as a formalization layer for my AI agents to help eliminate any ambiguity from my requirements. I really helps to have an intermediate layer between natural language specs and code. I have been struggling to make LLMs produce code that fulfills all my requirements and even skills like wayfinder and gri", "context": "", "use_case": "AI agent skills", "switched_from": "", "story": "An executable specification language based on temporal logic of actions (TLA)"}, {"id": 799173376, "item_id": 49948336, "url": "https://news.ycombinator.com/item?id=49948336", "by": "gregwebs", "date": "2026-10-03", "stance": "recommend", "quote": "I use mattpocock/skills and that generates ADRs (Architectural Decision Records).", "excerpt": "Agreed, and this seems better. My thought though has always been that I don't want there to be agent-only designated documentation. I use mattpocock/skills and that generates ADRs (Architectural Decision Records). That only uses skills, including a setup skill that will write a few pointers in AGENTS.md. I always have a CONTRIBUTING.md to document development flow and a CODING_STANDARDS.md. Between those and the READ", "context": "", "use_case": "generating ADRs for agents", "switched_from": "", "story": "Agents don't need memory, they need documentation"}, {"id": 798937235, "item_id": 49933577, "url": "https://news.ycombinator.com/item?id=49933577", "by": "nvarsj", "date": "2026-10-02", "stance": "recommend", "quote": "I basically just install pi-fabric, pi-blackhole, grill-me and that's it.", "excerpt": "Pi is the \"I use arch btw\" of harnesses right now. But I do love it - super customizable, and my token usage seems way more efficient (this is purely vibe based ofc) compared to codex/cc. I basically just install pi-fabric, pi-blackhole, grill-me and that's it. Tell my AGENTS.md to prefer subagent workflows. Herdr to manage all the terminals. I spend 99% of my professional work in this environment, I can't even remem", "context": "", "use_case": "Pi extension", "switched_from": "", "story": "Pi 1.0"}, {"id": 797846992, "item_id": 49865437, "url": "https://news.ycombinator.com/item?id=49865437", "by": "tadziokas", "date": "2026-09-27", "stance": "recommend", "quote": "I'll have to check out grill-me.", "excerpt": "I'll have to check out grill-me. It seems to slightly address a problem I have where for one off tasks it seems no workflow at all is necessary, but planning out projects before implementing them is rather unwieldy with claude especially when I move towards POC stages where the I feel like refactoring the design based on my PoC results is just an extremely detatched process.", "context": "", "use_case": "planning projects", "switched_from": "", "story": "Ask HN: What does your AI workflow look like?"}, {"id": 797845713, "item_id": 49865357, "url": "https://news.ycombinator.com/item?id=49865357", "by": "kingkongjaffa", "date": "2026-09-27", "stance": "recommend", "quote": "the highest ROI thing I've found is the concept of the grill-me skill", "excerpt": "- I have a central folder I use with context, project folders, tasks.md files etc. - I have a skill like /plan-my-day that makes a nice html file with tasks list, status of my work and projects etc. - when I'm starting up a new idea I use matt pocock's `grill-me` skill, I dictate a lot with https://handy.computer/ text going straight into claude code window. - once I have a decent brain dump I figure out what to do n", "context": "", "use_case": "refining new ideas", "switched_from": "", "story": "Ask HN: What does your AI workflow look like?"}, {"id": 797622720, "item_id": 49851420, "url": "https://news.ycombinator.com/item?id=49851420", "by": "diegof79", "date": "2026-09-25", "stance": "recommend", "quote": "The grill-me skill is much better for reviewing and clarifying choices", "excerpt": "Lately, I've been using Matt Pocock's \u201c/grill-me\u201d skill more and more, rather than plan mode or a complex, spec-driven set of skills. I made a personal fork of the skill to use Claude\u2019s ask tool, which has a nicer UX. I used plan mode for two reasons: to review the choices before execution, and to execute with another model (i.e., using the barely documented opusplan feature). The grill-me skill is much better for re", "context": "", "use_case": "reviewing and clarifying plans", "switched_from": "plan mode", "story": "Plan mode is dead"}, {"id": 792584640, "item_id": 49536540, "url": "https://news.ycombinator.com/item?id=49536540", "by": "Supermancho", "date": "2026-09-02", "stance": "warn", "quote": "The prose he writes often has very little utility, wrapped up in a ball of wax.", "excerpt": "Aside from the fact that Matt Pocock spends a lot of time with LLMS (only Claude really), he has a hand-wavy approach to engineering in a silo. He then uses that output as a foundation for other tools, while touting them as the best way to do things on youtube. Claude promotes this kind of developer and it's harmful. Anyone who uses LLMs notices (intrinsically or through investigation) that there are a small number o", "context": "", "use_case": "AI coding agent skills", "switched_from": "", "story": "AI Coding Agent Skills for Real Engineers"}, {"id": 789896145, "item_id": 49368509, "url": "https://news.ycombinator.com/item?id=49368509", "by": "mtzaldo", "date": "2026-08-19", "stance": "recommend", "quote": "Also, a skill like grill-me from Matt P. https://github.com/mattpocock/skills.", "excerpt": "It seems to me he could have use an skill like using-agent-skills from https://github.com/addyosmani/agent-skills go generate the specs and use a validator like oracle or something along the same lines. Also, a skill like grill-me from Matt P. https://github.com/mattpocock/skills.", "context": "", "use_case": "agent skill for refining specs", "switched_from": "", "story": "Sol Loves to Cheat"}, {"id": 788849056, "item_id": 49303066, "url": "https://news.ycombinator.com/item?id=49303066", "by": "zamalek", "date": "2026-08-14", "stance": "recommend", "quote": "In general, I find that the grill-me prompt[1] helps with this", "excerpt": "> Try as you might, it's nearly impossible to get the entirety of the context, intentions, business implications, budget constraints, and what-have-you written down and accessible to a coding agent. There will invariably be ambiguity and choices to be made, and it is nice to know that an agent will stop and ask when needed. In general, I find that the grill-me prompt[1] helps with this - but I am definitely not hand-", "context": "", "use_case": "clarifying context for coding agents", "switched_from": "", "story": "Why does Opus 5 feel worse to work with?"}, {"id": 788759040, "item_id": 49297440, "url": "https://news.ycombinator.com/item?id=49297440", "by": "gregwebs", "date": "2026-08-14", "stance": "recommend", "quote": "The grilling (grill-with-docs) skills [1] are amazing for ensuring you produce a through spec that covers all the edge cases.", "excerpt": "If you start with a spec you understand at the beginning then you don't need the LLM to generate high-level information about the changes at review time. The grilling (grill-with-docs) skills [1] are amazing for ensuring you produce a through spec that covers all the edge cases. The /code-review skill from there helps ensure that the code changes meet the spec. I use an intermediate detailed plan stage (done by a mor", "context": "", "use_case": "spec writing and code review", "switched_from": "", "story": "Understanding Is the New Bottleneck"}, {"id": 784974432, "item_id": 49060902, "url": "https://news.ycombinator.com/item?id=49060902", "by": "theptip", "date": "2026-07-26", "stance": "recommend", "quote": "I have been using https://www.aihero.dev/skills-grill-me with great success.", "excerpt": "I have been using https://www.aihero.dev/skills-grill-me with great success. It\u2019s a very nice small skill for lightweight up-front planning. And if you have that at the top of your session, you can just say \u201cplease use AskUserQuestion for any important decisions that arise which might alter our plan\u201d. It seems to prime the agent to be in high-level discourse mode.", "context": "", "use_case": "lightweight up-front planning", "switched_from": "", "story": "Claude Opus 5"}, {"id": 782021986, "item_id": 48876374, "url": "https://news.ycombinator.com/item?id=48876374", "by": "g42gregory", "date": "2026-07-11", "stance": "neutral", "quote": "I use Opus 4.8 with OMP/Pi coding agent and Matt Pocock Skills installed.", "excerpt": "I am ignoring Fable at the moment. It\u2019s on and off available, twice the price and does not seem to be better. I use Opus 4.8 with OMP/Pi coding agent and Matt Pocock Skills installed. I use professional/polite/questions-based communication pattern with Opus and it seems to work fine for coding. I am always aware that I need to justify my requests so it doesn\u2019t barf. Of course, I would never use Claude for anything cu", "context": "", "use_case": "agent skills", "switched_from": "", "story": "I used to love Claude, but the latest models are slowly ruining it"}, {"id": 781548304, "item_id": 48846769, "url": "https://news.ycombinator.com/item?id=48846769", "by": "Imanari", "date": "2026-07-09", "stance": "recommend", "quote": "Matt Pocock\u2019s \u201cgrill-me\u201d skill is a nice example of this idea", "excerpt": "I think a Q&A-based approach is could also be a suitable way to capture the reasoning behind a project. Instead of writing documentation afterward, an AI could interview the developer throughout the process and preserve the questions and answers as project context. Matt Pocock\u2019s \u201cgrill-me\u201d skill is a nice example of this idea: https://github.com/mattpocock/skills/tree/main/skills/productivity/grill-me", "context": "", "use_case": "capturing project reasoning via Q&A", "switched_from": "", "story": "How Version Control Will Evolve for the Agent Boom"}, {"id": 781038032, "item_id": 48814877, "url": "https://news.ycombinator.com/item?id=48814877", "by": "pipes", "date": "2026-07-07", "stance": "recommend", "quote": "Two skills I rely on are Matt pococks grill-me and his tdd. These massively helped", "excerpt": "Thanks. I have been doing this tdd abd it's a big improvement but the code is still pretty awful. Two skills I rely on are Matt pococks grill-me and his tdd. These massively helped, and technically the code generated is correct but it's still hard to follow and bloated. https://github.com/mattpocock/skills/blob/main/skills/productivity/grill-me/SKILL.md", "context": "", "use_case": "agent TDD workflow", "switched_from": "", "story": "Mark Zuckerberg tells staff that AI agents haven't progressed enough"}, {"id": 775091105, "item_id": 48443194, "url": "https://news.ycombinator.com/item?id=48443194", "by": "kingkongjaffa", "date": "2026-06-08", "stance": "recommend", "quote": "I've recently experimented with more lightweight things like https://github.com/mattpocock/skills  which are good.", "excerpt": "Claude skills made by other people are typically useless. The exceptions I have found are https://github.com/EveryInc/compound-engineering-plugin which was like an early brainstorm -> plan -> write -> embed knowledge and best practices. Which is a common workflow now. I've recently experimented with more lightweight things like https://github.com/mattpocock/skills which are good. Most work is just the same 'ask quest", "context": "", "use_case": "Claude skills", "switched_from": "", "story": "I'm tired of LLM skill slop, so I built mine with regression tests"}, {"id": 767848737, "item_id": 47990546, "url": "https://news.ycombinator.com/item?id=47990546", "by": "itissid", "date": "2026-05-02", "stance": "recommend", "quote": "the grill me skill(or its variation) at the design stage", "excerpt": "So RPI/QRSPI like skills (e.g. https://github.com/mattpocock/skills and https://github.com/humanlayer/humanlayer/tree/main/.claude/commands and https://github.com/dfrysinger/qrspi-plus ) for working with claude code work well enough for me that they can reliably* produce code that matches the plan/spec in a way they did not till December 2025. I have a gut feeling that these models can do just as well, has someone ru", "context": "", "use_case": "design-stage planning skills", "switched_from": "", "story": "DeepSeek V4\u2013almost on the frontier, a fraction of the price"}]}