How do I install the Grill With Docs skill?

Run npx skillmds@latest add navikt/grill-with-docs-2 in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Grill With Docs safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Grill With Docs?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Grill With Docs free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Grill With Docs?

navikt (@navikt) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Grill With Docs by navikt — Agent Skill

Name: Grill With Docs
Author: navikt

Grill With Docs

Resolve open requirements, assumptions or design choices against the repository's domain model, recording agreed terms and qualifying decisions. Use for coherent design clarification; use `to-spec` to capture a settled plan and `wayfinder` for dependent decisions needing a persistent map.

navikt Updated Sep 17, 2026

File contents

Grill with Docs

OpenCode v1: Skill names below are exact IDs from the active catalog, not slash commands. Load them with the native skill tool. Slash commands are direct user entry points only.

Load grilling and domain-modeling through the active skill catalog and the runtime's native loading mechanism. Combine the shared interview method with the repository's vocabulary and decision history. Start automatically when a request needs coherent design clarification; the user need not name this skill or separately select a documentation mode.

Inspect the relevant domain documents and implementation before asking. Test the proposal against established terms, boundaries, invariants, and concrete edge cases. Ask one material question at a time, recommend an answer with its consequence, and wait for the user's choice before dependent work.

As terms and decisions are resolved, apply domain-modeling within the user's authorized scope and the repository's language, paths, and formats. Write only what qualifies: a conversation can complete without creating a glossary entry or an ADR. Never turn an unchosen recommendation into a documented decision. Honor explicit discussion-only or read-only requests.

If the work exposes several dependent decisions that need a durable shared map across sessions, use wayfinder to organize them and continue this method for the active decision. Implementation size alone does not justify a map. When the direction is settled, summarize it and continue the authorized delivery workflow without an extra skill-selection or confirmation step.

Frequently asked questions