ace-pack 0.1.0

package/scripts/ace-project-presets.mjs

@@ -0,0 +1,158 @@
+export const DEFAULT_THRESHOLDS = {
+  large: {
+    minDiffLines: 300,
+    minFiles: 8,
+  },
+  small: {
+    maxDiffLines: 80,
+    maxFiles: 2,
+  },
+}
+
+export const UNIVERSAL_HIGH_RISK_PATHS = [
+  { label: 'agent instructions', pattern: 'AGENTS.md', tier: 'large' },
+  { label: 'agent instructions', pattern: 'CLAUDE.md', tier: 'large' },
+  { label: 'agent memory workflow', pattern: '.ai/**', tier: 'large' },
+  { label: 'CI workflow', pattern: '.github/workflows/**', tier: 'large' },
+  { label: 'CI workflow', pattern: '.gitlab-ci.yml', tier: 'large' },
+  { label: 'deployment configuration', pattern: 'Dockerfile', tier: 'large' },
+  { label: 'deployment configuration', pattern: 'docker-compose*', tier: 'large' },
+  { label: 'deployment configuration', pattern: 'compose.y*ml', tier: 'large' },
+  { label: 'automation scripts', pattern: 'scripts/**', tier: 'large' },
+  { label: 'request middleware', pattern: '**/middleware.*', tier: 'large' },
+  { label: 'database migration', pattern: '**/migrations/**', tier: 'large' },
+  { label: 'secret handling', pattern: '**/*secret*', tier: 'large' },
+  { label: 'token handling', pattern: '**/*token*', tier: 'large' },
+  { label: 'token or crypto handling', pattern: '**/*crypto*', tier: 'large' },
+  { label: 'authentication', pattern: '**/*auth*', tier: 'large' },
+]
+
+export const UNIVERSAL_HIGH_RISK_KEYWORDS = [
+  { keyword: 'auth', label: 'authentication', tier: 'large' },
+  { keyword: 'authentication', label: 'authentication', tier: 'large' },
+  { keyword: 'permission', label: 'permissions', tier: 'large' },
+  { keyword: 'permissions', label: 'permissions', tier: 'large' },
+  { keyword: 'secret', label: 'secret handling', tier: 'large' },
+  { keyword: 'secrets', label: 'secret handling', tier: 'large' },
+  { keyword: 'credential', label: 'credential handling', tier: 'large' },
+  { keyword: 'credentials', label: 'credential handling', tier: 'large' },
+  { keyword: 'password', label: 'password handling', tier: 'large' },
+  { keyword: 'passwords', label: 'password handling', tier: 'large' },
+  { keyword: 'token', label: 'token handling', tier: 'large' },
+  { keyword: 'tokens', label: 'token handling', tier: 'large' },
+  { keyword: 'crypto', label: 'token or crypto handling', tier: 'large' },
+  { keyword: 'migration', label: 'database migration', tier: 'large' },
+  { keyword: 'migrations', label: 'database migration', tier: 'large' },
+  { keyword: 'environment', label: 'environment isolation', tier: 'large' },
+  { keyword: 'production', label: 'production safety', tier: 'large' },
+  { keyword: 'deploy', label: 'deployment', tier: 'large' },
+  { keyword: 'delete', label: 'destructive operation', tier: 'standard' },
+  { keyword: 'destroy', label: 'destructive operation', tier: 'standard' },
+]
+
+const NEXT_TRPC_DRIZZLE_PATHS = [
+  { label: 'auth package', pattern: 'packages/auth/**', tier: 'large' },
+  { label: 'database schema', pattern: 'packages/db/src/schema/**', tier: 'large' },
+  { label: 'database migration', pattern: 'packages/db/migrations/**', tier: 'large' },
+  { label: 'database migration', pattern: 'packages/db/drizzle/**', tier: 'large' },
+  { label: 'tRPC router', pattern: 'packages/api/src/routers/**', tier: 'large' },
+  { label: 'Next.js middleware', pattern: 'middleware.ts', tier: 'large' },
+  { label: 'Next.js middleware', pattern: '**/middleware.ts', tier: 'large' },
+  { label: 'tenant isolation', pattern: '**/*tenant*', tier: 'large' },
+]
+
+const NEXT_TRPC_DRIZZLE_KEYWORDS = [
+  { keyword: 'tenant', label: 'tenant isolation', tier: 'large' },
+  { keyword: 'tenancy', label: 'tenant isolation', tier: 'large' },
+]
+
+export const PROJECT_PRESETS = {
+  'next-trpc-drizzle-saas': {
+    description: 'Next.js + tRPC + Drizzle SaaS monorepo risk profile.',
+    highRiskKeywords: dedupeKeywordRules([
+      ...UNIVERSAL_HIGH_RISK_KEYWORDS,
+      ...NEXT_TRPC_DRIZZLE_KEYWORDS,
+    ]),
+    highRiskPaths: dedupePathRules([...UNIVERSAL_HIGH_RISK_PATHS, ...NEXT_TRPC_DRIZZLE_PATHS]),
+    id: 'next-trpc-drizzle-saas',
+    name: 'Next/tRPC/Drizzle SaaS',
+  },
+}
+
+export const NEUTRAL_MEMORY_CONFIG = buildMemoryConfig({
+  highRiskKeywords: UNIVERSAL_HIGH_RISK_KEYWORDS,
+  highRiskPaths: UNIVERSAL_HIGH_RISK_PATHS,
+  profile: {
+    note: 'Run pnpm ace:onboard to profile this repository and apply project-specific risk rules.',
+    status: 'unprofiled',
+  },
+})
+
+export function getProjectPreset(id) {
+  return PROJECT_PRESETS[id] ?? null
+}
+
+export function buildPresetMemoryConfig(id, profile = {}) {
+  const preset = getProjectPreset(id)
+
+  if (!preset) {
+    throw new Error(`Unknown ACE project preset: ${id}`)
+  }
+
+  return buildMemoryConfig({
+    highRiskKeywords: preset.highRiskKeywords,
+    highRiskPaths: preset.highRiskPaths,
+    profile: {
+      preset: id,
+      status: 'profiled',
+      ...profile,
+    },
+  })
+}
+
+export function buildMemoryConfig({ highRiskKeywords, highRiskPaths, profile }) {
+  return {
+    _name: 'ACE (Agentic Context Engine) Configuration',
+    _profile: profile,
+    highRiskKeywords: dedupeKeywordRules(highRiskKeywords),
+    highRiskPaths: dedupePathRules(highRiskPaths),
+    thresholds: DEFAULT_THRESHOLDS,
+    version: 1,
+  }
+}
+
+export function dedupePathRules(rules) {
+  const seen = new Set()
+  const result = []
+
+  for (const rule of rules) {
+    const key = `${rule.pattern}:${rule.tier}:${rule.label}`
+
+    if (seen.has(key)) {
+      continue
+    }
+
+    seen.add(key)
+    result.push(rule)
+  }
+
+  return result
+}
+
+export function dedupeKeywordRules(rules) {
+  const seen = new Set()
+  const result = []
+
+  for (const rule of rules) {
+    const key = `${rule.keyword}:${rule.tier}:${rule.label}`
+
+    if (seen.has(key)) {
+      continue
+    }
+
+    seen.add(key)
+    result.push(rule)
+  }
+
+  return result
+}

package/scripts/ace-universal-doc-templates.mjs

@@ -0,0 +1,40 @@
+export const productRoadmapTemplate = `# Product Roadmap
+
+Use this file as the project-agnostic business context source for ACE.
+Keep entries concise and update it only when business goals, completed epics,
+or planned features materially change.
+
+## Business Goals
+- [Describe the active product/business goals]
+
+## Completed Epics
+- [Record completed epics with short outcome notes]
+
+## Planned Features
+- [Record planned features or roadmap bets]
+
+## Open Product Questions
+- [Record unresolved business or scope questions]
+`
+
+export const techDocsTemplate = `# Technical Docs
+
+Use this file as the project-agnostic technical context source for ACE.
+Keep it high-signal so browser-based AI architects can understand the system
+without reading large implementation files.
+
+## Architecture
+- [Summarize the current system architecture and key boundaries]
+
+## Data Model / DB Schema
+- [Summarize durable schema concepts, migrations, and data ownership]
+
+## Auth, RBAC, and Security
+- [Summarize auth flow, role boundaries, permissions, and token handling]
+
+## External APIs and Integrations
+- [Summarize external systems, APIs, sync jobs, and failure modes]
+
+## DevOps and Quality Gates
+- [Summarize setup, deployment, validation, and operational constraints]
+`

package/scripts/agent-memory-lib.mjs

@@ -0,0 +1,141 @@
+import path from 'node:path'
+
+import {
+  AGENTS_WORKFLOW_END_MARKER,
+  AGENTS_WORKFLOW_HEADER,
+  AGENTS_WORKFLOW_MARKER,
+  AI_FILE_SPECS,
+  CLAUDE_REQUIRED_SNIPPETS,
+  agentsWorkflowSection,
+  claudeTemplate,
+} from './agent-memory-templates.mjs'
+import { readTextIfExists, writeTextFile } from './ai-memory-utils.mjs'
+
+export {
+  AGENTS_WORKFLOW_END_MARKER,
+  AGENTS_WORKFLOW_HEADER,
+  AGENTS_WORKFLOW_MARKER,
+  AI_FILE_SPECS,
+  CLAUDE_REQUIRED_SNIPPETS,
+}
+
+async function ensureFile(filePath, content) {
+  const existingContent = await readTextIfExists(filePath)
+
+  if (existingContent !== null) {
+    return false
+  }
+
+  await writeTextFile(filePath, content)
+  return true
+}
+
+async function ensureAgentsWorkflow(filePath) {
+  const currentContent = await readTextIfExists(filePath)
+
+  if (currentContent === null) {
+    throw new Error(`Missing AGENTS.md at ${filePath}`)
+  }
+
+  if (currentContent.includes(AGENTS_WORKFLOW_MARKER)) {
+    if (!currentContent.includes(AGENTS_WORKFLOW_HEADER)) {
+      const upgradedContent = replaceMarkedSection(currentContent, agentsWorkflowSection)
+
+      if (upgradedContent !== currentContent) {
+        await writeTextFile(filePath, upgradedContent)
+        return true
+      }
+    }
+
+    return false
+  }
+
+  const nextContent = `${currentContent.trimEnd()}\n\n${agentsWorkflowSection}`
+  await writeTextFile(filePath, nextContent)
+  return true
+}
+
+function replaceMarkedSection(content, replacement) {
+  const startIndex = content.indexOf(AGENTS_WORKFLOW_MARKER)
+  const endIndex = content.indexOf(AGENTS_WORKFLOW_END_MARKER, startIndex)
+
+  if (startIndex === -1 || endIndex === -1) {
+    return content
+  }
+
+  const afterEndIndex = endIndex + AGENTS_WORKFLOW_END_MARKER.length
+
+  return `${content.slice(0, startIndex)}${replacement}${content.slice(afterEndIndex)}`
+}
+
+export async function ensureAgentMemory(rootDir) {
+  const createdFiles = []
+  const updatedFiles = []
+  const agentsPath = path.join(rootDir, 'AGENTS.md')
+  const claudePath = path.join(rootDir, 'CLAUDE.md')
+
+  if (await ensureAgentsWorkflow(agentsPath)) {
+    updatedFiles.push('AGENTS.md')
+  }
+
+  if (await ensureFile(claudePath, claudeTemplate)) {
+    createdFiles.push('CLAUDE.md')
+  }
+
+  for (const fileSpec of AI_FILE_SPECS) {
+    const absolutePath = path.join(rootDir, fileSpec.path)
+
+    if (await ensureFile(absolutePath, fileSpec.template)) {
+      createdFiles.push(fileSpec.path)
+    }
+  }
+
+  return {
+    createdFiles,
+    updatedFiles,
+  }
+}
+
+export async function validateAgentMemory(rootDir) {
+  const issues = []
+  const agentsPath = path.join(rootDir, 'AGENTS.md')
+  const claudePath = path.join(rootDir, 'CLAUDE.md')
+  const agentsContent = await readTextIfExists(agentsPath)
+  const claudeContent = await readTextIfExists(claudePath)
+
+  if (agentsContent === null) {
+    issues.push('Missing AGENTS.md')
+  } else if (!agentsContent.includes(AGENTS_WORKFLOW_MARKER)) {
+    issues.push(`AGENTS.md is missing the ${AGENTS_WORKFLOW_HEADER} section`)
+  } else if (!agentsContent.includes(AGENTS_WORKFLOW_HEADER)) {
+    issues.push(`AGENTS.md is missing the ${AGENTS_WORKFLOW_HEADER} section`)
+  }
+
+  if (claudeContent === null) {
+    issues.push('Missing CLAUDE.md')
+  } else {
+    for (const snippet of CLAUDE_REQUIRED_SNIPPETS) {
+      if (!claudeContent.includes(snippet)) {
+        issues.push(`CLAUDE.md is missing required content: ${snippet}`)
+      }
+    }
+  }
+
+  for (const fileSpec of AI_FILE_SPECS) {
+    const absolutePath = path.join(rootDir, fileSpec.path)
+    const content = await readTextIfExists(absolutePath)
+
+    if (content === null) {
+      issues.push(`Missing ${fileSpec.path}`)
+      continue
+    }
+
+    for (const snippet of fileSpec.requiredSnippets) {
+      if (!content.includes(snippet)) {
+        issues.push(`${fileSpec.path} is missing required content: ${snippet}`)
+      }
+    }
+  }
+
+  return issues
+}

package/scripts/agent-memory-templates.mjs

@@ -0,0 +1,350 @@
+import { NEUTRAL_MEMORY_CONFIG } from './ace-project-presets.mjs'
+import { productRoadmapTemplate, techDocsTemplate } from './ace-universal-doc-templates.mjs'
+
+export const AGENTS_WORKFLOW_HEADER = '## ACE (Agentic Context Engine) Workflow'
+export const AGENTS_WORKFLOW_MARKER = '<!-- agent-memory-workflow:start -->'
+export const AGENTS_WORKFLOW_END_MARKER = '<!-- agent-memory-workflow:end -->'
+export const CLAUDE_REQUIRED_SNIPPETS = ['AGENTS.md', '.ai/current-task.md']
+
+export const memoryConfigTemplate = JSON.stringify(NEUTRAL_MEMORY_CONFIG, null, 2)
+
+export const currentTaskTemplate = `# Current Task
+
+## Feature Name
+[Set the active feature or task name]
+
+## Lifecycle
+Status: active
+Version: v1
+Task Tier: standard
+Design Review Required: no
+Started: [YYYY-MM-DD HH:mm]
+Ready For Archive: no
+
+## Goal
+[Describe what is being built or changed]
+
+## Business Value / Product Alignment
+[Explain in 1-2 sentences why this matters to users or the business]
+
+## Technical Approach
+Option 1:
+- [Describe one viable approach and tradeoffs]
+
+Option 2:
+- [Describe another viable approach and tradeoffs]
+
+Chosen Approach:
+- [Explain why the selected approach best fits security, flexibility, architecture, and business value]
+
+## Current Status
+- [ ] Step 1
+- [ ] Step 2
+- [ ] Step 3
+
+## Affected Areas
+- [List the apps, packages, or files in scope]
+
+## Constraints
+- [List must-not-break rules or important boundaries]
+
+## Acceptance Criteria
+- [Describe the expected final behavior]
+
+## Completion Checklist
+- [ ] Goal completed
+- [ ] Acceptance criteria met
+- [ ] Tests/checks passed
+- [ ] \`.ai/session-handoff.md\` updated
+- [ ] \`.ai/changed-files.md\` updated
+- [ ] \`.ai/work-log.md\` updated
+- [ ] \`pnpm ace:validate\` passed
+- [ ] \`.ai/reflection-log.md\` updated if the task exposed friction or repeated mistakes
+- [ ] \`.ai/decisions.md\` updated if needed
+- [ ] \`.ai/tech-docs.md\` updated if architecture or technical state changed
+- [ ] \`.ai/product-roadmap.md\` updated if business or roadmap state changed
+- [ ] Final snapshot archived to \`.ai/archive/tasks/\` for large tasks
+- [ ] Next version/task defined if work continues
+`
+
+export const handoffTemplate = `# Session Handoff
+
+## Last Update
+[YYYY-MM-DD HH:mm]
+
+## What Was Done
+- [Summarize the latest completed work]
+
+## Current State
+- [Describe the current project or feature state]
+
+## Quality Review
+Product Alignment:
+- [Confirm the work still serves the stated business value]
+
+Architecture:
+- [Note the pattern used and why it fits the repo]
+
+Security:
+- [Note relevant auth, RBAC, tenancy, token, or data exposure risks]
+
+Code Quality:
+- [Note file size, duplication, strict typing, and test coverage risks]
+
+## Next Steps
+- [List the next concrete steps]
+
+## Known Issues
+- [List known gaps, risks, or blockers]
+
+## Verification
+- [List checks that passed or could not be run]
+
+## Notes
+- [Add any extra context another agent will need]
+`
+
+export const decisionsTemplate = `# Decisions
+
+## [YYYY-MM-DD HH:mm]
+
+Decision:
+- [Document the durable decision]
+
+Reason:
+- [Explain why it was made]
+
+Impact:
+- [Explain what this changes]
+`
+
+export const changedFilesTemplate = `# Changed Files
+
+[path/to/file]
+- [Short reason for the change]
+`
+
+export const workLogTemplate = `# Work Log
+
+## [YYYY-MM-DD HH:mm]
+
+- [Append short session summaries here]
+`
+
+export const reflectionLogTemplate = `# Reflection Log
+
+Use this file for short, actionable agent-process reflections. Do not log every
+minor task; record repeated tool friction, unclear prompts, poor assumptions,
+or workflow improvements worth carrying into future sessions.
+
+## Unresolved
+
+### [YYYY-MM-DD HH:mm] [Short issue title]
+Status: unresolved
+- Stuck Point: [Where the agent got stuck]
+- Likely Cause: [Tooling, prompt, missing context, or process issue]
+- Proposed Improvement: [Concrete change to try next time]
+
+## Resolved
+`
+
+export const claudeTemplate = `# CLAUDE.md
+
+## Repository Rules
+
+\`AGENTS.md\` is the authoritative source for repository constraints, stack
+choices, architecture rules, and quality gates. Read it first in every session.
+
+## Startup Workflow
+
+After reading \`AGENTS.md\`, read:
+
+1. \`.ai/report-brief.md\` if available for compact context.
+2. \`.ai/current-task.md\`
+3. \`.ai/session-handoff.md\`
+4. \`.ai/decisions.md\`
+5. \`.ai/changed-files.md\`
+6. Recent unresolved entries in \`.ai/reflection-log.md\`
+
+Read \`.ai/work-log.md\` only when you need additional history for the current
+task.
+
+## Working Rules
+
+- Prefer minimal diffs over rewrites.
+- Preserve existing TypeScript, UI, and API contracts unless the task says
+  otherwise.
+- Use \`pnpm ace:classify\` to select the adaptive task tier.
+- Run \`pnpm ace:onboard\` after fresh installation in an unfamiliar project
+  before trusting project-specific risk rules.
+- For large or high-risk standard tasks, complete \`.ai/current-task.md\`
+  Business Value and Technical Approach before writing code.
+- Treat \`.ai/*\` as the current source of task context and handoff state.
+- Use \`YYYY-MM-DD HH:mm\` timestamps in \`.ai/session-handoff.md\`,
+  \`.ai/work-log.md\`, \`.ai/reflection-log.md\`, and \`.ai/decisions.md\`.
+
+## End-of-Task Routine
+
+Run \`pnpm ace:validate\`, fix failures, then run \`pnpm ace:finish\` after
+updating the relevant \`.ai/*\` files. The finish script validates the
+required closeout depth for the detected task tier and generates the
+appropriate reports.
+`
+
+export const agentsWorkflowSection = `<!-- agent-memory-workflow:start -->
+## ACE (Agentic Context Engine) Workflow
+
+ACE is the local automation framework for managing AI context, code quality,
+and decision history. Use this workflow on top of the repo rules above;
+\`AGENTS.md\` remains the authoritative source for stack, architecture, and
+quality constraints.
+
+Before starting work:
+
+1. Read \`AGENTS.md\` first.
+2. If available, read \`.ai/report-brief.md\` first for a compact summary,
+   including recent unresolved reflections.
+3. Treat \`.ai/*\` as authoritative and read \`.ai/current-task.md\`,
+   \`.ai/session-handoff.md\`, \`.ai/decisions.md\`, and
+   \`.ai/changed-files.md\` when you need detail or verification.
+4. Run \`pnpm ace:classify\` before implementation to identify whether the
+   task is small, standard, or large.
+5. If this is a newly installed or unknown project and \`.ai/memory-config.json\`
+   is still marked \`unprofiled\`, run \`pnpm ace:onboard\` and apply an
+   approved profile before implementation.
+6. For large tasks, and standard tasks with high-risk signals, complete the
+   \`.ai/current-task.md\` Business Value and Technical Approach sections before
+   writing code. Compare at least two viable patterns and choose explicitly.
+7. Read \`.ai/work-log.md\` only when you need extra historical context.
+8. If the memory files are missing, run \`pnpm ace:init\`.
+
+Legacy commands such as \`pnpm ai:task:classify\`, \`pnpm ai:task:finish\`,
+and \`pnpm agent-memory:init\` remain supported for compatibility.
+
+While working:
+
+- Prefer minimal, safe diffs that preserve existing UI and API contracts.
+- Do not rewrite large components or architecture unless the task requires it.
+- Keep \`.ai/current-task.md\` aligned with the active task when scope changes.
+- Keep project-specific tier and risk rules in \`.ai/memory-config.json\`, the
+  canonical ACE config, not inside the scripts, so the toolset remains
+  portable.
+- Use \`pnpm ace:onboard\` to generate \`.ai/project-profile.md\` and
+  recommended project-specific risk rules when ACE is installed into an
+  unfamiliar repo.
+- When updating \`.ai/session-handoff.md\`, \`.ai/work-log.md\`,
+  \`.ai/reflection-log.md\`, or \`.ai/decisions.md\`, use timestamps in
+  \`YYYY-MM-DD HH:mm\` format.
+- Keep \`.ai/current-task.md\`, \`.ai/session-handoff.md\`,
+  \`.ai/reflection-log.md\`, and \`.ai/changed-files.md\` compact.
+- Archive only \`.ai/work-log.md\`, \`.ai/reflection-log.md\`, and
+  \`.ai/decisions.md\` into \`.ai/archive/\` when they grow past the documented
+  thresholds.
+- Use \`.ai/current-task.md\` lifecycle fields for task/version transitions.
+  When a large task version is complete, mark its completion checklist and let
+  \`pnpm ace:finish\` archive a final snapshot.
+
+After completing a task:
+
+1. Update the authoritative \`.ai/*\` files directly or through
+   \`ai:update:*\` helpers.
+2. Run \`pnpm ace:validate\` and fix any mechanical quality gate failures.
+3. Run \`pnpm ace:finish\` to validate the adaptive closeout and generate
+   reports.
+4. Small tasks need compact handoff, changed-files, work-log, and brief report.
+5. Standard tasks also need product, architecture, security, and code-quality
+   review notes.
+6. Large tasks also need design review, reflection entry when useful, archive
+   snapshot, full report, and a review of \`.ai/tech-docs.md\` or
+   \`.ai/product-roadmap.md\` when technical or business state changed.
+<!-- agent-memory-workflow:end -->
+`
+
+export const AI_FILE_SPECS = [
+  {
+    path: '.ai/memory-config.json',
+    requiredSnippets: [
+      '"_name": "ACE (Agentic Context Engine) Configuration"',
+      '"version": 1',
+      '"thresholds"',
+      '"highRiskPaths"',
+    ],
+    template: memoryConfigTemplate,
+  },
+  {
+    path: '.ai/current-task.md',
+    requiredSnippets: [
+      '# Current Task',
+      '## Lifecycle',
+      'Task Tier:',
+      'Design Review Required:',
+      '## Goal',
+      '## Business Value / Product Alignment',
+      '## Technical Approach',
+      '## Acceptance Criteria',
+      '## Completion Checklist',
+    ],
+    template: currentTaskTemplate,
+  },
+  {
+    path: '.ai/session-handoff.md',
+    requiredSnippets: [
+      '# Session Handoff',
+      '## What Was Done',
+      '## Quality Review',
+      '## Next Steps',
+    ],
+    template: handoffTemplate,
+  },
+  {
+    path: '.ai/decisions.md',
+    requiredSnippets: ['# Decisions', 'Decision:', 'Impact:'],
+    template: decisionsTemplate,
+  },
+  {
+    path: '.ai/product-roadmap.md',
+    requiredSnippets: [
+      '# Product Roadmap',
+      '## Business Goals',
+      '## Completed Epics',
+      '## Planned Features',
+    ],
+    template: productRoadmapTemplate,
+  },
+  {
+    path: '.ai/tech-docs.md',
+    requiredSnippets: [
+      '# Technical Docs',
+      '## Architecture',
+      '## Data Model / DB Schema',
+      '## Auth, RBAC, and Security',
+      '## External APIs and Integrations',
+    ],
+    template: techDocsTemplate,
+  },
+  {
+    path: '.ai/changed-files.md',
+    requiredSnippets: ['# Changed Files'],
+    template: changedFilesTemplate,
+  },
+  {
+    path: '.ai/work-log.md',
+    requiredSnippets: ['# Work Log'],
+    template: workLogTemplate,
+  },
+  {
+    path: '.ai/reflection-log.md',
+    requiredSnippets: ['# Reflection Log', '## Unresolved', '## Resolved'],
+    template: reflectionLogTemplate,
+  },
+  {
+    path: '.ai/archive/.gitkeep',
+    requiredSnippets: [],
+    template: '',
+  },
+  {
+    path: '.ai/archive/tasks/.gitkeep',
+    requiredSnippets: [],
+    template: '',
+  },
+]