accelint-onboard-openspec

Onboard OpenSpec

Guide the user through a conversational interview to produce a complete, project-specific openspec/config.yaml configured for the QRSPI methodology.

NEVER Do When Onboarding OpenSpec

• NEVER run codebase inference serially when subagents are available — Phase 3 spawns parallel subagents for different discovery domains. Serial scanning wastes time on codebases with many config files spread across directories. Spawn all 4 discovery agents simultaneously.

Companion Skill

This skill produces the project DNA layer of the agent instruction stack: structural facts about what the project is. It is the companion to the accelint-onboard-agents skill, which produces the behavior layer (AGENTS.md / CLAUDE.md): how the agent acts, communicates, and makes decisions.

If during this interview the user volunteers behavioral content (commit conventions, workflow steps, decision heuristics, tool preferences), acknowledge it and redirect: "That's behavioral — it belongs in AGENTS.md. I'll note it here for reference, but the accelint-onboard-agents skill is the right place to capture it." Do not write behavioral content into config.yaml.

AGENTS.md / CLAUDE.md  → accelint-onboard-agents skill  → HOW the agent behaves
openspec/config.yaml   → this skill             → WHAT the project is

---

Mental Model

The config has two jobs:

1. context: — Objective facts about the codebase injected into every AI artifact. Think of it as the "DNA" that makes AI suggestions feel native to the project. Facts only, no opinions.
2. rules: — Per-artifact checkpoints (proposal / design / tasks / spec) that encode the team's quality bar.

Phases

Phase 0 — File State Detection

Before any interview question is asked, check whether openspec/config.yaml exists and assess its state. Never silently pick a mode — always announce the detected mode to the user and confirm before proceeding.

Step 1 — Check for Related Documents

Before detecting config.yaml state, check for related onboarding documents:

1. Check for ARCHITECTURE.md
   • If exists: Read it to understand deployment and infrastructure
   • Use it to pre-fill answers for Turn 2 (infrastructure/deployment questions)
   • Note its existence for the "Related Documentation" section
   • Announce: "Found ARCHITECTURE.md — I'll use it to avoid asking questions about deployment that are already documented."

Note: AGENTS.md and README.md should NOT influence config.yml generation since they contain behavioral/usage info, not project DNA.

Step 2 — Detect Config State

After checking related documents, assess the config file state:

Does openspec/config.yaml exist?
│
├── No → MODE 1: Create
│         Full interview from scratch.
│
└── Yes → Read the file, then assess:
          │
          ├── Empty or near-blank (schema: line only, no context/rules)?
          │     → MODE 1: Create (with overwrite confirmation)
          │       Ask: "config.yaml exists but appears empty — should I
          │       populate it from scratch, or preserve any current content?"
          │
          ├── Contains recognised fields?
          │   (context: block present, rules: block with known artifact keys)
          │     → MODE 3: Refresh
          │       Abbreviated interview covering only detected drift and
          │       unresolved # TODO: fill in markers.
          │
          └── Contains real content in an unrecognised shape?
                → MODE 2: Import
                  Present three options (A / B / C) before proceeding.

Recognised shape = file is valid YAML with at least a context: key whose value is a non-empty string, or a rules: key with at least one of the known artifact IDs (proposal, specs, design, tasks).

---

Mode 1: Create

Run the full Phase 1 → Phase 2 → Phase 3 → Phase 4 interview. This is the happy path for a fresh repo.

---

Mode 2: Import

The file has real content that was not generated by this skill. Present the user with three options before touching anything:

"This config.yaml has existing content with a structure I don't recognise. How would you like to proceed?

(a) Restructure — I'll import your existing content, map it onto the context: / rules: schema, flag any material that belongs in AGENTS.md instead (workflow steps, commit conventions, tool preferences), run a targeted interview to fill gaps, and produce a merged file ready to replace the current one.

(b) Append — I'll run the full interview and add the skill's context: and rules: sections alongside your existing content without modifying what's already there.

(c) Dry run — I'll run the full interview and show you exactly what I would have generated, with no changes to the filesystem. Use this to evaluate fit before committing."

If option (a) is chosen:

1. Read the file in full.
2. Map existing content onto context: sub-sections and rules: artifact keys where possible.
3. Flag any content that violates the separation-of-concerns boundary (e.g., commit conventions, workflow steps, tool preferences, agent decision heuristics) — these belong in AGENTS.md. For each violation, ask: "This looks behavioral — it belongs in AGENTS.md. Should I move it there and remove it from config.yaml?"
4. Run a targeted interview covering only the gaps (context sub-sections with no existing coverage; artifact keys with no rules).
5. Show a merged preview before writing. Existing content is labelled # from existing file; new content is labelled # new.

If option (b) is chosen: Run the full Phase 1 → Phase 4 interview and write the generated context: and rules: blocks alongside existing content. Add a comment at the top: # Sections below added by accelint-onboard-openspec skill.

If option (c) is chosen: Run the full Phase 1 → Phase 4 interview and present the output in the conversation. Explicitly state: "No files were changed." Offer to re-run as (a) or (b) if the user is satisfied.

---

Mode 3: Refresh

The file matches the skill's expected schema — it was likely produced by a previous run. Run an abbreviated interview covering only:

1. Extract external findings — check if the invoking prompt includes a findings: list:

   • Parse the prompt for a findings: section (a bulleted list of factual statements)
   • Each finding is phrased as something already known to be true, never as an instruction
   • Example: "config.yaml's Anti-Patterns section says to avoid polling, but two archived changes chose polling for stated reasons"
   • Store these findings for merging in step 4

2. Drift detection — scan the codebase for changes since the file was last updated:

   Signal	Where to look
   Runtime / Node version changed	.nvmrc, .node-version, Dockerfile
   New packages / frameworks added	package.json deps, workspace roots
   TypeScript config tightened	tsconfig.json — new strict* flags
   New packages in monorepo	pnpm-workspace.yaml, turbo.json
   Build tooling changed	vite.config.*, tsup.config.*
   CI/CD workflows added	.github/workflows/
   New domain concepts	New top-level directories, new entity types in source
   Anti-patterns deprecated	@deprecated tags, // TODO: replace comments added

3. Unresolved TODOs — find all # TODO: fill in markers left from the previous run and surface them as targeted questions.

4. Merge and announce all findings before asking anything:

   • Combine external findings (from step 1) with drift findings (from step 2) and TODOs (from step 3)
   • Present the merged list to the user:

     "I found [N] external findings, [M] context sections that may have drifted, and [P] unresolved TODOs. I'll only ask about those — the rest looks current."

   • If external findings exist, note their source (e.g., "from completed OpenSpec change")

5. After the targeted interview, show only the changed sections in the preview before writing. Do not re-emit unchanged sections.

---

Phase 1 — Discovery Interview

Run the interview conversationally. Don't dump all questions at once. Group them into natural topic turns. If the user mentions a stack, infer related tooling and confirm rather than asking again.

Turn 1 — Project Identity

• What is the project name and its primary purpose?
• Monorepo, single package, or something else? If monorepo, what workspaces?
• Build system / task orchestration? (Turbo, Nx, Make, npm scripts, Makefile…)
• Package manager and any private registries? (npm, pnpm, yarn, bun…)

Turn 2 — Tech Stack (ask as a grouped block, not one by one)

• Runtime and version (Node.js 20, Bun 1.x, Python 3.12, etc.)
• Language + config (TypeScript strict? exactOptionalPropertyTypes? Python type hints?)
• Framework(s) and version (React 18, Next.js 14, Express, FastAPI, etc.)
• Key domain libraries (Deck.gl, Apache Arrow, Prisma, SQLAlchemy, etc.)
• Data layer (Postgres, MongoDB, DynamoDB, ORM/query builder, data formats)
• Testing setup (Vitest, Jest, Pytest, testing-library, Playwright, etc.)
• Linting / formatting (ESLint, Biome, Prettier, Black, Ruff, etc.)
• Build tools (Vite, tsup, esbuild, Webpack, etc.)
• CI/CD (GitHub Actions, CircleCI, etc.)
• Versioning approach (Changesets, standard-version, conventional commits, etc.)

Turn 3 — Architecture

• How is the codebase organised? (feature-based, layer-based, domain-driven?)
• Where does shared/utility code live?
• Any path aliases? (@/, ~/, src/, #lib/, etc.)
• Design patterns commonly in use? (factory, repository, observer, CQRS, etc.)

Turn 4 — Domain Concepts

• What are the 3–5 most important domain entities? Example prompt: "For a mapping app this might be Layer, Source, Viewport, Feature, Style."
• Any domain-specific terminology the AI should know?
• Any specialised concepts with non-obvious meanings in this codebase? Example: "orchestration" means something specific to us — it's the runtime layer that merges style with data, not a general workflow term.

Turn 5 — Performance

• Any concrete performance targets? (p95 < 200 ms, 60 fps, < 50 MB heap, etc.)
• Known hot paths or performance-critical areas?
• Memory or bundle-size constraints?

Turn 6 — Code Patterns

• Export style: named exports, default exports, or mixed?
• Naming conventions: files, variables, functions, constants? Example: "kebab-case files, camelCase vars, SCREAMING_SNAKE_CASE for constants, PascalCase for types."
• Error handling: throw, Result<T,E>, error boundaries, something else?
• Testing structure: describe/it, test/expect, AAA pattern?
• Test file location: co-located with source or a separate __tests__/ tree?
• Fixture / factory approach for test data?

Note: Commit message convention is a workflow procedure — it belongs in AGENTS.md, not here. If the user raises it now, capture it mentally and surface it in the accelint-onboard-agents skill. Do not add it to config.yaml.

Turn 7 — Anti-Patterns

• Any patterns explicitly banned in code review?
• Deprecated patterns still in the codebase that new code should NOT emulate?
• Known performance traps specific to this stack?

Turn 8 — Proposal Rules What does YOUR team require in a proposal? Good prompts:

• "Do you need proposals to call out database migration impact?"
• "Do you need proposals to flag API breaking changes?"
• "Any security review checklist items?"

Turn 9 — Design Rules Project-specific design concerns to encode? Good prompts:

• "Docker / Kubernetes resource changes to document?"
• "Performance implications section required?"
• "Specific architecture diagram style (ASCII, Mermaid)?"

Turn 10 — Task Rules

• How do you tag tasks by package or module? Example: [PKG:auth], [MODULE:pipeline], GitHub labels…
• Rollback plan required for database changes?
• Deployment-specific test gates (smoke tests, canary checks)?

---

Phase 2 — Smart Defaults

After each stack answer, surface relevant conventions to confirm. Use these examples as a pattern; extend to other stacks as appropriate.

Next.js + TypeScript + Tailwind → suggest confirming:

• App Router vs Pages Router and which patterns apply
• Server Component vs Client Component boundary rules
• "use client" directive placement convention
• API route organisation (app/api/ vs pages/api/)

React + Vitest + testing-library → suggest confirming:

• userEvent over fireEvent preference
• screen query priority (role > label > testid)
• render wrapper for providers

Python + FastAPI → suggest confirming:

• Pydantic v1 vs v2 (different field-validator syntax)
• Dependency injection for DB sessions (Depends)
• Alembic migration workflow
• lifespan vs startup/shutdown event hooks

Node.js + Prisma → suggest confirming:

• prisma.$transaction patterns
• Soft-delete vs hard-delete convention
• Migration naming convention

---

Phase 3 — Parallel Codebase Inference

After the interview, spawn parallel discovery subagents to fill remaining config gaps. All config sections are load-bearing — a missing field degrades every downstream AI artifact, so inference is always preferable to omission.

Spawn discovery subagents in parallel — don't scan serially. Each agent focuses on one inference domain and returns structured findings. Wait for all agents to complete, then merge results before Phase 4.

Spawn these agents simultaneously:

Agent A — Stack & Build Tooling

• Runtime / Node version: .nvmrc, .node-version, package.json#engines, Dockerfile
• TypeScript config: tsconfig.json (compilerOptions flags, paths aliases)
• Package manager: package-lock.json, yarn.lock, pnpm-lock.yaml, bun.lockb
• Monorepo workspaces: package.json#workspaces, pnpm-workspace.yaml, turbo.json, nx.json
• Build tools: vite.config.*, webpack.config.*, tsup.config.*, esbuild scripts
• Return: runtime version, TS config flags, package manager, workspace list, build tools

Agent B — Testing & Code Quality

• Test framework: vitest.config.*, jest.config.*, pytest.ini, pyproject.toml#tool.pytest
• Linting / formatting: .eslintrc*, biome.json, .prettierrc*, ruff.toml
• Test structure: Sample test files — describe/it nesting depth, file location relative to source
• Test file type checking: CI scripts, package.json — check if tsc --noEmit runs on *.test.ts files
• Property-based testing: Check for fast-check in dependencies
• Vitest mock cleanup: vitest.config.ts — check for clearMocks, mockReset, restoreMocks
• Return: test framework, code quality tools, test structure patterns, type checking config

Agent C — Architecture & Code Patterns

• Architecture organisation: Directory tree of src/ or workspace roots — infer feature-based vs layer-based
• Path aliases: tsconfig.json#compilerOptions.paths, vite.config#resolve.alias
• Design patterns: Sample source files — look for factory functions, repository objects, observer hooks
• Export style: Sample 3–5 source files; tally named vs default exports
• Naming conventions: Sample file names, exported identifiers; describe what you observe
• Error handling: Grep for throw, Result, Either, tryCatch, error boundary components
• TypeScript baseline patterns: If tsconfig.json exists, flag that TS/JS baseline patterns should be included
• Return: architecture style, path aliases, design patterns, export conventions, naming patterns, error handling approach

Agent D — CI/CD & Versioning

• CI/CD: .github/workflows/, .circleci/, Jenkinsfile
• Versioning: .changeset/, CHANGELOG.md, commitlint.config.*, .releaserc*
• Anti-patterns: eslint rule overrides marked off or warn, comments like // TODO: replace, @deprecated
• Return: CI/CD platform, versioning approach, documented anti-patterns

After all agents complete: merge their findings into a unified inference map. Tag each field as INFERRED [source] or UNKNOWN. Fields tagged UNKNOWN should be marked as # TODO: fill in in the config preview.

For each field resolved via inference, note the source in the preview with a trailing comment, e.g.:

- Runtime: Node.js 20 LTS   # inferred from .nvmrc
- Language: TypeScript 5.4, strict, exactOptionalPropertyTypes   # inferred from tsconfig.json

If a field genuinely cannot be inferred (e.g., performance targets, domain concepts, team-specific rules), mark it with # TODO: fill in rather than omitting it. The user can resolve these after reviewing the preview. Do not silently drop a section — an explicit TODO is a prompt to act; an absent section is an invisible gap.

---

Phase 4 — Generation

1. Show a labeled preview of the full config before writing anything. Inferred values carry their source comment; unresolved fields carry # TODO: fill in. This gives the user a complete picture of confidence level across every field.
2. Ask: "Does this look right? Any sections to correct or expand before I write the file?"
3. After confirmation, write to openspec/config.yaml (create directory if needed), stripping the inference source comments — they are for review only, not the final file. For the Related Documentation section: only include links to files that actually exist in the repository. Check for each file (ARCHITECTURE.md, AGENTS.md/CLAUDE.md, README.md) before including its link.
4. Validate the generated YAML — after writing, read the file back and verify:
   • No tabs (YAML requires spaces for indentation)
   • Values with special characters are properly quoted
   • No syntax errors (unmatched brackets, quotes, etc.)
   • The file can be conceptually parsed as valid YAML If validation reveals issues, fix them immediately and rewrite the file.
5. Print a brief summary of what was configured, what was inferred vs answered directly, and which # TODO fields still need human input.

---

YAML Generation Safety Rules

CRITICAL: YAML syntax is strict about special characters. Follow these rules when generating config.yaml to avoid syntax errors:

Quoting Requirements

Rule: Values that start with special YAML characters need quoting.

Special characters: |, >, ", ', (, ), [, ], {, }, *, &, !, %, @, `

Examples:

# Parentheses at start of value
❌ description: (internal) auth module     # Syntax error
✅ description: "(internal) auth module"   # Quoted

# Square brackets (looks like YAML list syntax)
❌ tag: [PKG:auth]                         # YAML thinks it's a list
✅ tag: "[PKG:auth]"                       # Quoted string

# Pipe character (YAML thinks it's block scalar)
❌ pattern: some|other                     # Syntax error
✅ pattern: "some|other"                   # Quoted

# Colon in value (YAML thinks it's a nested key)
❌ note: Time: 5pm                         # Syntax error
✅ note: "Time: 5pm"                       # Quoted

# Value containing quotes - escape with opposite quote type
✅ command: 'npm run "test:unit"'          # Single quotes protect doubles
✅ command: "npm run 'test:unit'"          # Double quotes protect singles

Multi-line String Handling

Use block scalar indicators for multi-line content:

# Literal block (preserves newlines) - preferred for context field
context: |
  Line 1
  Line 2
  Line 3

# Folded block (folds newlines into spaces) - rarely needed
description: >
  This is a long
  description that
  flows together.

Indentation Rules

• Use spaces only — never tabs
• Consistent indent — typically 2 spaces per level
• Block scalars — content inside | or > must be indented relative to the key

Rules for List Values

# Simple list items - no quotes needed for plain text
rules:
  proposal:
    - Keep proposals under 100 lines
    - Include scope boundaries

# List items with special chars - quote them
rules:
  tasks:
    - "Tag with [PKG:name] format"        # Quotes protect [ and ]
    - 'Use "Test:" prefix for validation' # Single quotes protect inner "

Validation Checklist

After generating the config, mentally verify:

1. No bare (, ), |, ", ' immediately after colons (unless using | or > for multiline)
2. No tab characters anywhere in the file
3. Consistent 2-space indentation throughout
4. All list items (-) aligned at the same indent level within their parent
5. Quoted strings use matching quote types

If any of these rules are violated, the YAML will fail to parse.

---

Config Template

Use this exact structure. Fill every [placeholder] with content from the interview or codebase inference. If a field cannot be resolved by either means, replace its placeholder with # TODO: fill in — never omit the field. Every section is load-bearing for downstream AI artifact quality.

schema: spec-driven

# Project Context
# Injected into every AI-generated artifact (proposal, design, spec, tasks).
# QRSPI principle: objective research layer — facts only, no opinions.

context: |
  # ═══════════════════════════════════════════════════════════════════════════
  # STACK FACTS
  # ═══════════════════════════════════════════════════════════════════════════

  ## Project Identity
  [project name and one-sentence purpose]
  [repo structure: monorepo / single-package / workspaces list]
  [build system and task orchestration]
  [package manager + registries]

  ## Tech Stack
  - Runtime:            [e.g., Node.js 20 LTS]
  - Language:           [e.g., TypeScript 5.4, strict mode, exactOptionalPropertyTypes]
  - Framework:          [e.g., Next.js 14 App Router]
  - Key Libraries:      [domain-specific dependencies with versions]
  - Data Layer:         [databases, ORMs, data formats, query builders]
  - Testing:            [framework, utilities, coverage tooling]
  - Linting/Formatting: [tools and config files in use]
  - Build Tools:        [bundlers, compilers, transpilers]
  - CI/CD:              [platform and key workflow names]
  - Versioning:         [release strategy and changelog tooling]

  ## Architecture Patterns
  - Organisation: [feature-based / layer-based / domain-driven / other]
  - Shared code:  [path to shared utilities / packages]
  - Path aliases: [list of aliases and their resolved paths]
  - Key patterns: [design patterns in common use]

  ## Domain Concepts
  - [Entity or concept]: [one-line definition]
  - [Entity or concept]: [one-line definition]
  - [Entity or concept]: [one-line definition]

  ## Performance Targets
  - [metric]: [target value and context]

  ### TypeScript/JavaScript Performance (if applicable)
  - Hot paths:    [functions executed >1000 times per interaction or >100 times/sec]
  - Frame budget: [for real-time systems: 60fps = 16.67ms, 120fps = 8.33ms]
  - Constraints:  Bounded iteration (explicit limits on loops/queues), O(n) or better algorithmic complexity

  # ═══════════════════════════════════════════════════════════════════════════
  # PATTERNS TO FOLLOW
  # ═══════════════════════════════════════════════════════════════════════════

  ## Code Patterns
  - Exports:         [named / default / mixed — and when each applies]
  - Naming:          [files, variables, functions, constants, types]
  - Error handling:  [throw / Result<T,E> / boundaries / other]
  - Validation:      [approach and library]
  - Constants:       Use `as const` objects, never `enum`
  - Classes:         Prefer functions over classes unless state management required or extending existing class
  - Return values:   Return zero values (empty array, empty string, 0, false) instead of null/undefined
  - Leaf functions:  Leaf functions (bottom of call stack) should be pure — same inputs produce same outputs, no side effects. Centralize state manipulation in parent/orchestrator functions.
  - Type safety:     Avoid `any` (use `unknown` or generics); avoid `enum` (use `as const` objects); use `type` over `interface`
  - Immutability:    Prefer `const`, immutable data structures, pure functions
  - Documentation:   Comprehensive JSDoc for all exported code (@param, @returns, @template, @example)
  - Order:           Internal functions, variables and types should be defined before they are used (internal/export types -> internal/export constants -> internal/export functions)
  - Parameter order: Data-last ordering — place the data being operated on as the final parameter. Enables partial application and composition.
  - Composition:     Use curried functions when the same first parameter(s) recur across call sites.

  ## Architecture Patterns
  - [pattern name]: [brief description of how it's used here]

  ## Testing Patterns
  - Pattern:        AAA (Arrange, Act, Assert) with clear boundaries
  - Property-based: (If available) Use `fast-check` for encode/decode pairs, validators, normalizers, pure functions
  - Test scope:     Never test library internals; never export internals to test them; never mock own pure functions
  - Structure:      [describe/it nesting convention]
  - File location:  [co-located / __tests__ / other]
  - Test doubles:   Hierarchy: real implementation > fakes > stubs > spies > mocks
  - Fixtures:       [factory functions / fixture files / inline data]
  - Assertions:     [preferred assertion style]
  - Nesting:        Max 2 levels of describe blocks — use descriptive test names instead
  - Verification:   MUST run `tsc --noEmit` on test files before marking complete
  - Benchmarks:     [approach if any]

  # NOTE: Commit message convention, PR workflow, and tool preferences
  # are behavioral — they belong in AGENTS.md, not here.

  # ═══════════════════════════════════════════════════════════════════════════
  # PATTERNS TO AVOID
  # ═══════════════════════════════════════════════════════════════════════════

  ## Code Anti-Patterns
  - Using `any` instead of `unknown` or generics
  - Using `enum` instead of `as const` objects
  - Using `interface` when `type` works (prefer type)
  - Returning `null`/`undefined` instead of zero values (empty arrays, empty strings, 0, false)
  - Not validating external data with schemas
  - Deep nesting instead of early returns

  - [anti-pattern]: [why it's banned or deprecated]

  ## Performance Anti-Patterns
  - Chaining array methods (`.filter().map().reduce()`) — use single reduce pass
  - Using `Array.includes()` for repeated lookups (use `Set.has()` for O(1) lookups)
  - Recomputing constants inside loops (hoist invariants outside)
  - Unbounded loops or queues (set explicit limits to prevent runaway resource consumption)
  - Placing `try/catch` in hot paths (V8 cannot inline, 3-5x slowdown)

  - [anti-pattern]: [why it's banned or deprecated]

  ## Testing Anti-Patterns
  - Testing library internals (e.g., verifying Array.prototype.map works)
  - Exporting internal functions just to test them
  - Loose assertions in tests (toBeTruthy, toBeDefined)
  - Nested describe blocks >2 levels deep
  - Testing implementation details instead of behavior

  - [anti-pattern]: [why it's banned or deprecated]

  ## Documentation Anti-Patterns
  - Missing JSDoc on exported functions/types
  - Documenting HOW instead of WHAT/WHY in JSDoc
  - Vague comment markers (`// TODO: fix this` instead of `// TODO: Replace with binary search for O(log n)`)

# ═══════════════════════════════════════════════════════════════════════════
# PER-ARTIFACT RULES
# ═══════════════════════════════════════════════════════════════════════════

rules:
  proposal:
    # QRSPI: Scope definition, not a plan.
    - State the requirement or ticket driving this change
    - Define scope boundaries — explicitly list what is OUT of scope
    - Keep under 100 lines (tight and focused)
    [user-specific proposal rules]

  design:
    # QRSPI: The "brain surgery" checkpoint — reviewed before any code is written.
    # Target ~200 lines capturing current state, desired state, open questions.

    # Required sections (in this order):
    - Start with "Current State": what the code does today, key files, entry
      points, relevant data flows
    - "Desired End State": what changes after this work, what stays the same
    - "Patterns to Follow": ONLY if specific files/functions to reference exist
      for this change's domain
    - "Patterns to Avoid": ONLY if specific anti-patterns apply to this change
    - "Open Questions": genuine uncertainties requiring human input. If none,
      state explicitly "No unresolved questions."
    - "Resolved Decisions": numbered (Decision 1, Decision 2…) with Choice,
      Rationale, Alternatives Considered

    # Technical depth:
    - Use ASCII diagrams for data flows, state machines, architecture
    - Call out performance implications where relevant
    [user-specific design rules]

    # Constraints:
    - Keep under 250 lines total

  tasks:
    # QRSPI: Vertical slicing for early failure detection.

    # Vertical slicing (strong preference):
    - Order as vertical slices — each task delivers a testable end-to-end path
    - Do NOT group by architectural layer unless explicitly justified
    - Horizontal (layer-by-layer) only for pure infrastructure; include
      justification in the task description when used
    - Each task MUST include an explicit "Test:" line describing what to verify
      before proceeding to the next task
    - Prefer 3–5 major slices; more than 5 suggests scope is too large

    # Granularity:
    - Max 2 hours per task; break larger work into subtasks
    [user-specific task tagging, e.g., [PKG:name] or [MODULE:name]]
    - Call out inter-task dependencies explicitly
    [user-specific rollback requirements]
    [user-specific deployment test gates]

  spec:
    - Use Given/When/Then for behaviour specifications
    - Include concrete example data relevant to the domain
    - Document edge cases explicitly
    [user-specific spec rules]

# ═══════════════════════════════════════════════════════════════════════════
# RELATED DOCUMENTATION
# ═══════════════════════════════════════════════════════════════════════════
# Include only files that actually exist in the repository:
# - ARCHITECTURE.md: System overview, deployment, component interactions, data flows
# - AGENTS.md: Agent behavior rules, workflow procedures, communication style
# - README.md: Installation, quick start, usage guide

---

Interaction Principles

• Conversational, not interrogative. Bundle related questions into a single turn. Use natural language, not bullet-dump forms.
• Infer and confirm. "You mentioned Vitest — I'll assume you're using @testing-library/react for component tests; correct?" is better than asking from scratch.
• Examples reduce ambiguity. When asking about naming conventions, give an example first so the user can pattern-match.
• Iterative. Let the user amend answers. Don't lock them into the first response.
• Preview before writing. Always show the full generated config and get explicit confirmation before touching the filesystem.
• Infer before asking, ask before omitting. Always attempt codebase inference for any unanswered field. If inference fails, surface a # TODO rather than dropping the section. A config with explicit TODOs is actionable; a config with missing sections silently degrades every artifact it drives.