Architect — Proactive Configuration for New Projects

Purpose

Capture architectural intent before code exists, so AI tools enforce conventions from line one. This is the same AI configuration layer that

/bootstrap

produces (CLAUDE.md, rules, agents, AGENTS.md) — but declared from intent rather than discovered from code.

This is NOT scaffolding — no code is generated. Only AI tool configuration.

Lifecycle:

/architect     →  declares intent, generates [DECLARED] config
  write code   →  AI tools enforce declared conventions
/bootstrap     →  validates reality vs declarations
                   CONFIRMED → remove [DECLARED] tag
                   DIVERGED  → present both, ask user
                   NOT YET   → keep [DECLARED] tag
/learn         →  ongoing refinement

When to Use

• Starting a brand new project with no code yet
• When you know the tech stack and architecture but haven't written code
• When you want AI tools to enforce conventions from the first commit

Prerequisites

• A target directory (can be empty)
• The user has an idea of what they're building (at minimum: stack and project name)

Tool Target Preferences

Before generating any output, check for

.north-starr.json

in the project root:

{
  "version": 1,
  "targets": ["claude", "copilot", "cursor"]
}

• If the file exists, only generate artifacts for the listed targets
• If the file is missing, ask the user which tools they use (as part of Phase 1), save their answer to

  .north-starr.json

  , then generate only for those targets

• AGENTS.md

  is always generated regardless of preferences (it's universal)
• Valid targets:

  "claude"

  ,

  "copilot"

  ,

  "cursor"

Workflow

Phase 1: Identity (Required)

Goal: Capture the minimum viable project identity — enough for smart defaults.

Ask these questions one at a time, adapting based on answers:

1. Project name + description

   • "What's the project name and a one-sentence description?"
   • Accept anything — this becomes the header in context files

2. Tech stack

   • "What tech stack? (e.g., 'React TS', 'Go API', 'iOS SwiftUI', 'Rails', 'Python FastAPI')"
   • Accept shorthand — resolve to full stack details using

     references/stack-conventions.md

   • If ambiguous, offer the 2-3 most likely interpretations

3. Architecture pattern

   • Based on the chosen stack, propose 2-3 common patterns from

     references/stack-conventions.md

   • "For [stack], the most common patterns are [X, Y, Z]. Which fits, or describe your own?"
   • If the user isn't sure, suggest the default for that stack
   • Accept custom patterns — just document what the user describes

4. AI tools (only if

   .north-starr.json

   is missing)
   • "Which AI tools do you use? (Claude Code, VS Code Copilot, Cursor — or all?)"
   • Save the answer to

     .north-starr.json

     in the project root
   • If the file already exists, skip this question and use the saved preferences

After Phase 1: You have enough to generate useful configuration. Offer to continue to Phase 2 or generate now with smart defaults.

Phase 2: Conventions (Recommended)

Goal: Refine defaults — confirm or override what the stack conventions suggest.

4. File/directory organization

   • Propose the default layout from

     references/stack-conventions.md

     for the chosen stack+pattern
   • "Here's the typical structure for [stack] + [pattern]. Does this match your plan, or would you change anything?"
   • User can confirm, modify, or skip (defaults apply)

5. Conventions to enforce

   • Propose defaults from the stack conventions: naming, error handling, testing strategy, state management
   • "These are the standard conventions for [stack]. Any changes or additions?"
   • Focus on things that are enforceable via rules (not aspirational guidelines)

6. Hard constraints / danger zones

   • "Are there any hard rules — things that should never happen? (e.g., 'no direct DB access outside the repository layer', 'never force-unwrap optionals')"
   • These become high-priority rules

After Phase 2: Configuration is well-tailored. Offer Phase 3 or generate.

Phase 3: Context (Optional)

Goal: Capture extra context that improves AI effectiveness.

7. Reference project

   • "Is there an existing north-starr project to adapt config from?"
   • If yes, read that project's config and adapt (preserving the new project's identity)

8. Catch-all

   • "Anything else? Domain vocabulary, team preferences, external constraints, API contracts?"
   • Capture whatever the user offers

Phase 4: Generate Configuration

Goal: Produce all configuration artifacts, marked

[DECLARED]

to distinguish from discovered config.

Use

references/stack-conventions.md

to fill in smart defaults for anything the user didn't explicitly specify.

---

A. Project Context (root-level)

Write to enabled target locations:

• CLAUDE.md

  — Claude Code (auto-loaded) — generate if

  claude

  target is enabled

• AGENTS.md

  — Universal (works with any AI tool) — always generated

• .github/copilot-instructions.md

  — VS Code Copilot (auto-loaded) — generate if

  copilot

  target is enabled

All context files get the same content:

# [Project Name]

[One-sentence project description]

<!-- Generated by /architect — [DECLARED] config. Run /bootstrap after writing code to validate. -->

## Tech Stack

[Languages, frameworks, key dependencies, build tools, test runner]

## Architecture [DECLARED]

[Pattern, topology, layers and their responsibilities]

## Grain [DECLARED]

[What changes easily, what is hard — based on declared architecture]

## Module Map [DECLARED]

[Planned modules/packages with one-line descriptions and intended relationships]

## Vocabulary

[Domain terms with project-specific meanings]

## Conventions [DECLARED]

[Naming, error handling, testing, state management — from stack defaults + user overrides]

<!-- [NORTH-STARR:how-to-approach-tasks] -->
## How to Approach Tasks

**Before starting any task, assess its complexity:**

| Complexity | Signals | Action |
|------------|---------|--------|
| **Low** | Single file, straightforward change, well-understood area | Implement directly |
| **Medium** | 3+ files, unfamiliar area, non-trivial logic, or integration points | Run `/invert` then `/plan` automatically |
| **High** | Wide blast radius, irreversible changes, unfamiliar domain, or critical system | Run `/invert` then `/plan` automatically |

**For Medium and High complexity tasks — do this automatically, don't wait for the user to ask:**
1. Run `/invert` — identify risks, edge cases, and failure modes
2. Run `/plan` — break the work into tracked tasks, using invert's risks as constraints
3. Execute the plan with progress tracking and session notes

**Resuming work:** Always check `.plans/` for active implementation plans before starting new work. If an active plan exists, resume it instead of starting fresh.
<!-- [/NORTH-STARR:how-to-approach-tasks] -->

<!-- [NORTH-STARR:auto-learn] -->
## When to Learn Automatically

**Run `/learn` automatically — do not wait for the user to ask — when any of these signals occur during a session:**

| Signal | Example | What to Capture |
|--------|---------|-----------------|
| **User corrects your approach** | "No, don't do it that way — use X instead" | **Pattern** — the correct approach so it's followed next time |
| **Same fix requested twice** | User asks you to fix the same issue or area more than once in a session | **Landmine** — the fragile area and why it keeps breaking |
| **Your change breaks something** | Tests fail, build breaks, or existing behavior regresses after your edit | **Landmine** — what broke and why, so it's avoided next time |
| **User rejects generated code** | "That's wrong", "revert that", or user manually undoes your change | **Pattern or Landmine** — capture what was wrong and what's correct |
| **You discover an undocumented convention** | Code follows a pattern not captured in any rule or context file | **Pattern** — document it before it's forgotten |
| **You hit a trap not in any landmine rule** | Something looked safe but caused unexpected problems | **Landmine** — document the trap for future sessions |

**How auto-learn works:**
1. Detect the signal during normal work
2. Finish the immediate fix or correction first
3. Then run `/learn` to capture the insight as a pattern or landmine rule
4. If a pattern or landmine already exists for this area, update it — do not create duplicates. Prompt the user when the update contradicts existing content.
<!-- [/NORTH-STARR:auto-learn] -->

If any of these files already exist with project-specific content, merge rather than overwrite.

---

B. Path-Scoped Rules

Generate rules based on the declared architecture and conventions. Create for each enabled target:

Claude Code —

.claude/rules/*.md

(if

claude

target enabled):

---
paths: ["glob/pattern/**"]
---
<!-- Generated by /architect — [DECLARED] -->

[Clear, concise instruction based on declared conventions]

VS Code Copilot —

.github/instructions/*.instructions.md

(if

copilot

target enabled):

---
applyTo: "glob/pattern/**"
---
<!-- Generated by /architect — [DECLARED] -->

[Same instruction content]

Cursor —

.cursor/rules/*.mdc

(if

cursor

target enabled):

---
globs: glob/pattern/**
---
<!-- Generated by /architect — [DECLARED] -->

[Same instruction content]

What to generate rules for:

• Naming conventions — scoped to the project's file extensions
• Architecture layer constraints — what each layer can/cannot import
• Error handling — the declared pattern for the stack
• Testing conventions — naming, location, style
• Any hard constraints from Phase 2

Guidelines:

• Generate rules from declared conventions and stack defaults — not discovered code
• Use specific path globs based on the declared directory structure
• Keep each rule file focused on one concern
• Mark every rule with

  <!-- Generated by /architect — [DECLARED] -->

---

C. Agents

Generate agents for each enabled target that supports them:

Claude Code —

.claude/agents/*.md

(if

claude

target enabled):

---
name: [project]-explorer
description: Deep exploration and analysis of the [project] codebase
model: sonnet
tools: Read, Glob, Grep
memory: project
---

VS Code Copilot —

.github/agents/*.agent.md

(if

copilot

target enabled):

---
name: [project]-explorer
description: Deep exploration and analysis of the [project] codebase
tools: codebase
---

The agent prompt should include:

• The declared architecture and grain
• Planned modules and their relationships
• Declared conventions to watch for

---

D. Startup Plan (Optional)

Offer to generate

.plans/PLAN-project-setup.md

— an initial implementation plan that bridges into

/plan

:

# Plan: project-setup

**Created:** [date]
**Status:** ACTIVE
**Source:** /architect — initial project setup

## Goal
Set up the project structure, dependencies, and initial code based on the declared architecture.

## Tasks

### 1. Initialize project
**Status:** PENDING
- [ ] Create project with [stack's init tool]
- [ ] Configure [build tool, linter, formatter]
- [ ] Set up directory structure per declared module map

### 2. Implement foundational patterns
**Status:** PENDING
**Blocked by:** 1
- [ ] Create base types/interfaces for [architecture pattern]
- [ ] Set up [error handling, DI, state management] patterns
- [ ] Add first example following all declared conventions

### 3. Validate with /bootstrap
**Status:** PENDING
**Blocked by:** 2
- [ ] Run /bootstrap to validate declared config against actual code
- [ ] Review any DIVERGED items
- [ ] Confirm conventions are reflected in the codebase

Adapt the tasks to the specific stack and architecture.

Post-Architect Checklist

• .north-starr.json

  exists (created if missing during Phase 1)

• AGENTS.md

  at root (always)

• CLAUDE.md

  at root (if

  claude

  target enabled)

• .github/copilot-instructions.md

  (if

  copilot

  target enabled)
• Path-scoped rules in enabled tool formats
• At least one project-tuned agent per enabled tool that supports agents
• All sections marked

  [DECLARED]

• Startup plan offered

Output Summary

After completing all phases, present:

## Architect Complete

**Project:** [name]
**Tech Stack:** [languages, frameworks, tools]
**Architecture:** [pattern] [DECLARED]
**Grain:** [easy changes vs. hard changes] [DECLARED]
**Enabled Tools:** [list from .north-starr.json]

**Files Generated:**

Universal:
- AGENTS.md — [sections included]

[Include only sections for enabled targets:]

Claude Code:                              ← if claude target enabled
- CLAUDE.md — [sections included]
- [N] .claude/rules/ files — [list names]
- [N] .claude/agents/ files — [list names]

VS Code Copilot:                          ← if copilot target enabled
- .github/copilot-instructions.md
- [N] .github/instructions/ files — [list names]
- [N] .github/agents/ files — [list names]

Cursor:                                   ← if cursor target enabled
- [N] .cursor/rules/ files — [list names]

**All config marked [DECLARED].**
Run /bootstrap after writing code to validate declarations against reality.

Notes

• This skill is language-agnostic — it uses

  references/stack-conventions.md

  for smart defaults across 11+ stacks
• This skill respects tool target preferences — check

  .north-starr.json

  for enabled targets, or ask and save if missing. Only generate artifacts for enabled tools.
• Even answering only Phase 1 should produce useful, non-trivial configuration via smart stack-aware defaults
• The

  [DECLARED]

  tag is critical — it tells

  /bootstrap

  to validate rather than regenerate
• Conversational style — ask questions one at a time, adapt based on answers, offer smart defaults
• No code generation — only AI tool configuration
• Works best as a complement to

  /bootstrap

  : architect declares, bootstrap validates
• The generated configuration is a starting point — it improves through

  /bootstrap

  validation and

  /learn

  refinement