grill-with-docs · git:20260523.36e5a7e · 2026-05-23 · sha256 79a011fd5e9f7325
grill-with-docs git:20260523.36e5a7eA
Immutable. This exact content never changes and is served at /api/v1/blob/79a011fd5e9f7325.
No tool to install first
claude plugin marketplace add https://markdownregistry.com/mp/a/art_ddo5ye3pfsfw7f3p/79a011fd5e9f7325ce2409751b7c9dd067498fef1cb48416d206e35c876675d4.json
Then /plugin install grill-with-docs@markdownregistry-ddo5ye3pfsfw7f3p in Claude Code. It installs this exact version, the one graded A here, pinned to the commit it was graded at, and leaves it there.
--- name: grill-with-docs description: Grilling session that stress-tests a plan against this repo's documented language (glossary, conventions, explanation docs, ADRs) and updates those docs inline as decisions crystallise. Use when the user wants to pressure-test a design against the project's existing vocabulary and prior decisions. --- <what-to-do> Interview me relentlessly about every aspect of this plan until we reach a shared understanding. Walk down each branch of the design tree, resolving dependencies between decisions one-by-one. For each question, provide your recommended answer. Ask the questions one at a time, waiting for feedback on each question before continuing. If a question can be answered by exploring the codebase, explore the codebase instead. </what-to-do> <supporting-info> ## Where the "domain model" lives in this repo During codebase exploration, also look for existing documentation: | Source | What it covers | | :------------------------------- | :------------------------------------------------------------------------ | | `docs/README.md` (Glossary) | Cross-cutting nouns: Organization, Member, Session, Participant, Gateway, Provider, etc. | | `docs/conventions/naming.md` | Role taxonomy and naming stems for modules / classes | | `docs/explanation/<topic>.md` | Current narrative for a subsystem (system architecture, security model) | | `docs/adr/` | Dated log of why a hard-to-reverse choice was made | Before grilling, skim the Glossary, the relevant `docs/explanation/` doc, and any ADRs already filed for the area. For *why*/*how* questions that span multiple services, delegate to the `architecture-explainer` subagent rather than re-reading docs in the main context. ## During the session ### Challenge against the existing language When the user uses a term that conflicts with the Glossary or `naming.md`, call it out. Example: "Glossary defines `Session` as the live client connection context; you're using it for the engine process that runs it. Which do you mean?" ### Sharpen fuzzy language Propose precise canonical terms; pull from the existing Glossary first, only invent when nothing fits. Common ambiguities here: - "Session" (the user-facing connection context vs. the engine process running it) - "Message" (the inbound event vs. the rendered delivery to a Channel) - "Tenant" vs. `Organization` (BetterAuth term wins) - "Member" vs. "Participant" vs. "User" ### Stress-test with concrete scenarios Force precision with edge cases that touch service boundaries: - "What happens to an in-flight Message dispatch when the Participant disconnects mid-send?" - "If the Gateway and Session Engine disagree on a Session's active state, who wins?" - "A Provider for one Channel is down: does the Session block, drop, or fan out to the other Channels?" ### Cross-reference with code When the user states how something works, verify against the code in the relevant service (`apps/acme-api/`, `services/acme-session-engine/`, `services/acme-gateway/`, `apps/acme-web/`). Surface contradictions: "`session.manager.ts` tears down on the last Participant leaving, but you said Sessions stay warm. Which is right?" ### Update the existing docs inline When something resolves, update it in place. Capture as it happens; don't batch. - **New cross-cutting noun?** Add to the Glossary table in `docs/README.md`. - **Naming stem or role suffix decision?** Update `docs/conventions/naming.md`. - **Subsystem narrative has drifted from reality?** Update the relevant `docs/explanation/<topic>.md`. - **Hard-to-reverse choice with non-obvious rejected alternatives?** Open an ADR. Do not create a parallel `CONTEXT.md`. See [CONTEXT-FORMAT.md](./CONTEXT-FORMAT.md) for the underlying glossary discipline if you need a reminder of what a good entry looks like. ### Offer ADRs sparingly Only offer an ADR when all three are true: 1. **Hard to reverse**: the cost of changing your mind later is meaningful 2. **Surprising without context**: a future reader will wonder "why did they do it this way?" 3. **The result of a real trade-off**: there were genuine alternatives and one was picked for specific reasons If any of the three is missing, skip it. See [ADR-FORMAT.md](./ADR-FORMAT.md) and [docs/adr/README.md](../../../docs/adr/README.md) for the bar and template. </supporting-info>