grill-with-docs
Use when a plan must be stress-tested against the codebase's existing language and decisions — grill it against current docs, glossary terms, and ADRs before implementation begins
Best use case
grill-with-docs is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Use when a plan must be stress-tested against the codebase's existing language and decisions — grill it against current docs, glossary terms, and ADRs before implementation begins
Teams using grill-with-docs should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/grill-with-docs/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How grill-with-docs Compares
| Feature / Agent | grill-with-docs | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Use when a plan must be stress-tested against the codebase's existing language and decisions — grill it against current docs, glossary terms, and ADRs before implementation begins
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# Grill With Docs Grill With Docs is a documentation-grounded grilling session. It does the same branch-by-branch interrogation as `grill-me`, but every challenge is anchored in the project's current language, design records, and observable implementation. ## When to Use - A plan must align with existing docs, glossary terms, ADRs, or architecture notes - The same concept is described differently across code, docs, and the current proposal - You need to challenge a plan without drifting away from the repo's existing vocabulary - A durable decision may need to be recorded while the conversation is still active ## When NOT to Use | Instead of grill-with-docs | Use | |----------------------------|-----| | You only need plan interrogation, not documentation grounding | `grill-me` | | You are writing or updating docs after implementation changed | `doc-update` | | You need a broad codebase walkthrough or onboarding tour | `code-tour` | ## Workflow ### 1. Gather the current sources of truth Before questioning the plan, inspect the best available durable references: - root docs such as `README.md` or architecture notes - glossary or context files if they exist - ADRs or decision logs - the current code path if the docs are stale or incomplete If the repo does not maintain dedicated glossary or ADR files, use the nearest durable project docs instead of inventing new ceremony. ### 2. Challenge the plan against documented language When the plan uses a term that conflicts with current project language, call it out directly. Examples: - "The docs call this a workspace, but the plan says project. Which term is canonical?" - "The ADR says we prefer thin adapters here. Why does this proposal add a new orchestration layer?" ### 3. Use one question at a time Like `grill-me`, walk the tree branch by branch. Each question should either: - confirm the plan fits the existing docs - expose a contradiction - force a new durable decision ### 4. Check code when the docs and proposal disagree If the docs say one thing and the current code does another, surface the mismatch instead of assuming the docs win automatically. ```markdown **Documented:** feature flags gate this flow. **Observed in code:** the path is unconditional in production. **Question:** Which source should the new plan follow? ``` ### 5. Record only durable decisions If the session produces a decision worth preserving, update the relevant docs during the session. Only create or propose an ADR when all three are true: 1. it is hard to reverse 2. it would be surprising without context 3. it resulted from a real tradeoff If one of those is missing, leave it out of the ADR layer. ## Output Template ```markdown ## Doc-Grounded Grill Summary ### Confirmed by Existing Docs - ... ### Conflicts Found - ... ### Decisions Made - ... ### Docs To Update - ... ``` ## Common Rationalizations | Rationalization | Reality | |----------------|---------| | "The docs are probably stale, so ignore them." | Maybe — but you still need to surface the contradiction explicitly. | | "We'll rename the terms later." | Vocabulary drift makes design reviews and follow-up work harder immediately. | | "Let's write an ADR for everything." | ADRs are for durable, surprising, hard-to-reverse tradeoffs, not routine notes. | ## Red Flags - The plan uses new terminology without checking whether the repo already has a term - Code, docs, and proposal disagree, but the session never resolves which one is authoritative - ADRs are proposed for obvious or low-cost decisions - The grilling session turns into doc rewriting before the core plan is settled ## Verification - [ ] The plan was checked against current docs or code, not memory alone - [ ] Terminology conflicts were surfaced explicitly - [ ] Questions were asked one at a time - [ ] Only durable decisions were promoted into doc updates or ADR candidates ## See Also - [`grill-me`](../grill-me/SKILL.md) — one-question-at-a-time grilling without the doc-grounding requirement - [`doc-update`](../../documentation/doc-update/SKILL.md) — sync docs after implementation changes - [`architecture-decisions`](../../documentation/architecture-decisions/SKILL.md) — capture hard-to-reverse technical decisions deliberately
Related Skills
grill-me
Use when a plan still has hidden assumptions — run a one-question-at-a-time grilling session until dependencies, risks, and decisions are explicit before implementation starts
verification-before-completion
Use before claiming any task is done — run the exact command that proves the fix works, read the output, and only then report success.
using-git-worktrees
Use when you need multiple branches checked out at once — create isolated working directories for parallel development without cloning the repository repeatedly
triage
Use when a single issue needs structured triage — classify it, reproduce if needed, request missing information, and leave a durable brief or close-out note in the tracker.
to-issues
Use when a plan, spec, or PRD must become an actionable backlog — break it into thin dependency-aware issues that each deliver a verifiable vertical slice
sprint-workflow
Use when starting a new feature, refactor, or multi-step dev task — runs the full sprint cycle (Think → Plan → Build → Review → Test → Ship → Monitor) using Copilot CLI's plan/autopilot modes.
sprint-retro
Use at the end of a sprint to run a data-driven retrospective — analyzes session history and git metrics to surface what shipped, what slowed you down, and concrete improvements.
security-audit
Use when a codebase needs a formal security audit beyond a quick scan — applies OWASP Top 10 and STRIDE threat modeling from a CSO perspective to surface systemic vulnerabilities.
release
Use when a sprint or feature is complete and ready to ship — tags the version, generates GitHub Release notes, runs rollout or smoke verification, and publishes to npm/PyPI/Docker registries.
prompt-optimizer
Use when a rough prompt, vague idea, or task description needs to become a finished copy-pasteable prompt for a chat-based LLM - rewrite it into one ready-to-send prompt with no blanks, no placeholders, and a clear output shape.
outside-voice
Use when you need an independent second opinion before, during, or after implementation — run challenge, consult, or review mode in a direct builder-to-builder voice
llm-wiki
Use when research or domain knowledge keeps getting rediscovered across sessions — build a supplementary markdown wiki that compounds synthesized knowledge without replacing GitHub or committed project guidance