/grill-docs
Run the grilling interview and write each settled term and decision to the repo as it settles, so they survive compaction or an abandoned session.
$ npx -y skills add bendrucker/claude --skill grill-docs --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
- Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
- You can call itInvoke it directly when you want it.
- Slash command
/grill-docs
Context preview
The summary Claude sees to decide when to auto-load this skill.
Run the grilling interview and write each settled term and decision to the repo as it settles, so they survive compaction or an abandoned session.
SKILL.md
grill-docs.SKILL.mdname: grill-docs
description: Run the grilling interview and write each settled term and decision to the repo as it settles, so they survive compaction or an abandoned session.
argument-hint: "[<what to grill>]"
disable-model-invocation: true
allowed-tools:
- AskUserQuestion
- Agent
- Read
- Write
- Edit
- Glob
- Grep
Grill With Docs
Run the `grilling` skill's interview over $ARGUMENTS, and write each term and decision to disk as soon as it is settled.
Writing as we go is the point. If the session is compacted or abandoned, the repo still holds what we agreed.
Where It Lands
Choose both destinations before the first round and say them back to me.
Terms go where the repo already keeps instructions Claude loads automatically:
- A term that applies to one part of the repo goes in `.claude/rules/<area>.md` under a `paths` entry, so it loads when that part is touched.
- A term that applies everywhere goes in `CLAUDE.md`.
- Create the file if it does not exist.
Decisions go beside the plan that produced them, as `<plan>-decisions.md` in the same directory.
Vocabulary
Write a term as soon as it is settled, which means I have picked both the word and its meaning.
Prefer a word the codebase already uses. When the question is what to call something, search the codebase for candidates first and offer those as the options.
A definition says what the thing is and how it differs from the thing it is most easily confused with. Keep it to that.
Decisions
Record a decision when all three hold:
- Reversing it later would be expensive.
- Someone who was not in this conversation would not guess it.
- It was a tradeoff: a reasonable alternative was rejected.
Most sessions produce no decision that meets all three, and that is expected.
A record states the decision, the alternatives it rejects, and the tradeoff.
Close
Write the interview's closing escalation contract (what to proceed on without asking, what to stop and consult on) into the decisions file, then wait for me to confirm it.
Read more
name: grill-docs description: Run the grilling interview and write each settled term and decision to the repo as it settles, so they survive compaction or an abandoned session. argument-hint: "[<what to grill>]" disable-model-invocation: true allowed-tools: - AskUserQuestion - Agent - Read - Write - Edit - Glob - Grep
Grill With Docs
Run the `grilling` skill's interview over $ARGUMENTS, and write each term and decision to disk as soon as it is settled.
Writing as we go is the point. If the session is compacted or abandoned, the repo still holds what we agreed.
Where It Lands
Choose both destinations before the first round and say them back to me.
Terms go where the repo already keeps instructions Claude loads automatically:
- A term that applies to one part of the repo goes in `.claude/rules/<area>.md` under a `paths` entry, so it loads when that part is touched.
- A term that applies everywhere goes in `CLAUDE.md`.
- Create the file if it does not exist.
Decisions go beside the plan that produced them, as `<plan>-decisions.md` in the same directory.
Vocabulary
Write a term as soon as it is settled, which means I have picked both the word and its meaning.
Prefer a word the codebase already uses. When the question is what to call something, search the codebase for candidates first and offer those as the options.
A definition says what the thing is and how it differs from the thing it is most easily confused with. Keep it to that.
Decisions
Record a decision when all three hold:
- Reversing it later would be expensive.
- Someone who was not in this conversation would not guess it.
- It was a tradeoff: a reasonable alternative was rejected.
Most sessions produce no decision that meets all three, and that is expected.
A record states the decision, the alternatives it rejects, and the tradeoff.
Close
Write the interview's closing escalation contract (what to proceed on without asking, what to stop and consult on) into the decisions file, then wait for me to confirm it.
My personal plugin marketplace for Claude Code, Anthropic's AI coding assistant.
Repo: bendrucker/claude
Other skills on bendrucker-claude.
cleye
Type-safe CLI argument parsing with cleye, the standard parser for this repo's Bun scripts. Use when writing or editing any script that takes arguments (flags,…
activity
Report real device usage from ActivityWatch. Covers per-app time, window titles, and active vs idle spans. Use when asked "what apps did I use", "how long was…
history
Report shell history from atuin's local capture. Covers what commands ran, when, where, and how they exited. Use when asked "what commands did I run", "what…
bun
Bun runtime patterns. Use when running bun commands, working with package.json/bun.lock, writing TypeScript scripts under Bun, or developing Claude Code…
agent-team
Orchestrating Claude Code agent teams. Use when creating teams, spawning teammates, assigning tasks, configuring teammate modes, or setting up team quality…

