grill-me-plus
Skill0 starsChanged 32 days agoMITView on GitHub
Interview the user relentlessly about a plan or design until you reach shared understanding, resolving each branch of the decision tree and recording decisions to disk. Use when the user wants to stress-test a plan, pressure-test a design, or asks to be grilled.
In this repository10 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- archifySkill
- capture-compareSkill
- find-skillsSkill
- gated-tddSkill
- eli5Skill
- idea-bankingSkill
- jasoseo-coachSkill
- thinking-aloud-writingSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 10
If a question can be answered by exploring the codebase, explore it instead of asking.
- Distinctive: line 15
Always offer a third option: "I don't have enough knowledge to decide this." If I take it, don't let me rubber-stamp it: either teach me …
- Distinctive: line 27
When a later answer implies revising an earlier decision, flag it: "This implies revisiting decision #N.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Interview me relentlessly about every aspect of this plan until we reach shared understanding. Walk down each branch of the design tree, resolving dependencies one at a time. Ask one question per turn and wait for my answer.
If a question can be answered by exploring the codebase, explore it instead of asking.
For each question, give your recommended answer and your confidence. When there's an obvious default, recommend it confidently. When it's a genuine trade-off, lay out the options neutrally and argue the strongest case against your own pick. Always offer a third option: "I don't have enough knowledge to decide this." If I take it, don't let me rubber-stamp it: either teach me briefly, or mark the decision as an open spike to research later.
Order questions by risk, not by ease. Surface the hardest-to-reverse, highest-uncertainty decisions first. Don't linger on cheap, reversible leaves.
Maintain a DECISIONS.md log. Append each resolved decision immediately, especially hard-to-reverse ones; don't wait for a batch. After each append, show me a one-line diff of what changed so I can catch drift. Every ~8 decisions, re-read the whole log and reconcile it against our conversation.
When a later answer implies revising an earlier decision, flag it: "This implies revisiting decision #N. Revise now, or note it and defer?"
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.