About grill-with-docs-smarter
A personal skill for design discussions using `/grilling` and `/domain-modeling`. It resolves uncertainties from conversation, documentation, or bounded web research before asking questions that could change the recommendation.
Key features
- 01Filter questions by available evidence
- 02Separate inference, research, and user input
- 03Present recommendations and tradeoffs
- 04Keep web research bounded
- 05Ask one dependency-changing question at a time
Use cases
- 01Discuss design decisions in large codebases
- 02Assess current library or framework status
- 03Review security guidance, standards, or platform rules
- 04Maintain domain-modeling documentation
Install
Copy the command and run it in a terminal.
npx skills add https:/ / github.com/ gil-1/ skills --skill grill-with-docs-smarterAfter installing, type/grill-with-docs-smarterin Claude Code to confirm it loaded.
Details
From the project READMERun a /grilling session, using the /domain-modeling skill, with an evidence-first question filter.
Before asking the user, decide whether the uncertainty should be:
- Inferred from the conversation, repo, or existing docs.
- Checked with bounded web research because current external facts could change the recommendation.
- Resolved by giving the agent's recommended answer.
- Asked because only the user can answer it.
Do not ask questions answered by the first three. When evidence supports a direction, present the recommendation, tradeoff, and rejected alternatives before asking. Ask only questions whose answers would materially change the recommendation.
Use web research when uncertain, time-sensitive external facts could materially change the design: library or framework state, security guidance, standards, pricing, or platform rules. Keep it bounded; do not turn a simple grill into a deep research task unless the decision needs it.
Default to the simplest recommendation that satisfies the user's constraints. Complexity must earn its place by reducing real risk, preserving needed options, or matching scale, security, or product constraints.
Ask one dependency-changing question at a time. Group up to three independent questions only when answering them together is easier for the user.
Continue maintaining docs through /domain-modeling.