its-magic 0.1.2-9 → 0.1.3-0

package/README.md

@@ -1,755 +1,1602 @@
-# its-magic — AI dev team
-
-Drop-in template repo that implements a structured GSD workflow in Cursor:
-intake -> discovery -> architecture -> sprint plan -> execute -> QA -> release,
-with pause/resume, decision gates, and persistent artifacts.
-
-## Quick start
-
-1. Use `/gsd-intake` to capture the idea and kick off questions.
-2. Continue with `/gsd-discovery`, `/gsd-architecture`, `/gsd-sprint-plan`.
-3. Execute work with `/gsd-execute`, verify via `/gsd-qa`, finalize via
-   `/gsd-release`.
-4. Use `/gsd-pause` anytime and `/gsd-resume` to continue from artifacts.
-5. Use `/gsd-refresh-context` to compact state and prevent context rot.
-
-## How to set up a project with its-magic
-
-its-magic is an installer you run once per repo. It copies the AI dev team
-workflow files (`.cursor/` commands, rules, agents, hooks, skills, plus `docs/`,
-`sprints/`, `handoffs/`, etc.) into your project. After that, all `/gsd-*`
-commands are available in Cursor.
-
-### Step 1: Install its-magic (once)
-
-Pick **one** method — they all do the same thing:
-
-| Method | Install command |
-|--------|----------------|
-| npm    | `npm install -g its-magic` |
-| npx    | No install needed (runs on demand) |
-| Chocolatey | `choco install its-magic` (run as Admin) |
-| Homebrew | `brew tap USER/tap && brew install its-magic` |
-
-### Step 2: Set up a new or existing repo
-
-**New empty project:**
-
-```bash
-mkdir my-project && cd my-project
-git init
-its-magic --target . --mode missing --create
-```
-
-**Existing project (add workflow without touching existing files):**
-
-```bash
-cd my-existing-project
-its-magic --target . --mode missing
-```
-
-**Existing project (update/overwrite with latest workflow, backup first):**
-
-```bash
-cd my-existing-project
-its-magic --target . --mode overwrite --backup
-```
-
-If you use **npx** (no global install), prefix with `npx`:
-
-```bash
-npx its-magic --target . --mode missing
-```
-
-### Step 3: Open in Cursor and start working
-
-```
-1. Open the project folder in Cursor
-2. Type /gsd-intake and describe your idea
-3. Follow the workflow: discovery -> architecture -> sprint plan -> execute -> QA -> release
-```
-
-### What gets installed into your repo
-
-```
-your-project/
-  .cursor/
-    commands/       ← /gsd-* slash commands
-    rules/          ← agent behavior rules
-    agents/         ← subagent definitions
-    skills/         ← GSD team skill + templates
-    hooks/          ← safety & context hooks
-    hooks.json      ← hook configuration
-    scratchpad.md   ← automation flags
-  docs/             ← product & engineering docs
-  sprints/          ← sprint planning & tracking
-  handoffs/         ← role-to-role notes
-  decisions/        ← decision records
-  .github/workflows/ ← CI/CD templates
-```
-
-### Options reference
-
-| Flag | Description |
-|------|-------------|
-| `--target <path>` | Target repo path (required, or `.` for current dir) |
-| `--mode missing` | Only copy files that don't exist yet (safe default) |
-| `--mode overwrite` | Replace all files with latest versions |
-| `--mode interactive` | Prompt per file |
-| `--backup` | Save existing files to `gsd-backups/<timestamp>/` before overwriting |
-| `--create` | Create the target directory if it doesn't exist |
-| `--help` | Show usage info |
-
-## Repository layout (what is what)
-
-### GSD tooling (engine)
-- `.cursor/` — commands, rules, agents, hooks, skills, scratchpad config.
-- `.github/workflows/` — CI/CD templates driven by `docs/engineering/runbook.md`.
-- `gsd-installer.*` — installers for adding its-magic to other repos.
-- `scripts/` — helper scripts (release notes generator, unified release script).
-- `packaging/` — Homebrew formula + Chocolatey nuspec for multi-platform distribution.
-
-### GSD artifacts (project memory)
-- `docs/` — product + engineering docs (vision, architecture, decisions, state).
-- `sprints/` — sprint planning, tasks, progress, QA, UAT.
-- `handoffs/` — role-to-role handoff notes.
-- `decisions/` — decision records (DEC-xxxx).
-- `milestones/` — milestone tracking (optional).
-
-### Examples and code
-- `examples/` — reference apps created by benchmarks or demos.
-
-### Testing and benchmarks
-- `tests/` — test harness for its-magic verification.
-- `benchmarks/` — scenario, live, prompted, and headless benchmark runners.
-
-### Misc
-- `Plan.md` — original master plan for its-magic.
-- `.gitignore` — ignores generated reports, temp folders, and telemetry.
-
-## Voice input (multilingual)
-
-Voice is only an input layer. It produces text that feeds the same workflow:
-
-- Option A: OS dictation (no setup, language support varies by OS)
-- Option B: Cursor voice (if available)
-- Option C: Local STT (whisper / whisper.cpp style dictation)
-
-Recommended reliability pattern for slash commands:
-
-- Bind a text expander or hotkey to insert `/gsd-intake ` (or any command).
-- Dictate only the content after the command.
-
-## CI/CD via runbook
-
-Workflows in `.github/workflows/` read command keys from
-`docs/engineering/runbook.md`. If a key is not set (or is a placeholder),
-the workflow skips the step and exits successfully.
-
-Supported keys:
-
-- `TEST_COMMAND`
-- `LINT_COMMAND`
-- `TYPECHECK_COMMAND`
-- `DEPLOY_STAGING_COMMAND`
-- `DEPLOY_PROD_COMMAND`
-
-## Installer / updater
-
-Use one of the installers below to add its-magic to an existing repo or
-bootstrap an empty one:
-
-- Windows: `gsd-installer.ps1`
-- macOS/Linux: `gsd-installer.sh`
-- Python fallback: `gsd-installer.py`
-
-Modes:
-
-- `missing` copies only files that do not exist
-- `overwrite` replaces existing files
-- `interactive` prompts per file
-
-Backup option:
-
-- Use `--backup` (or choose when prompted) to save existing files into
-  `gsd-backups/<timestamp>/` before overwriting.
-
-Examples:
-
-- `powershell -ExecutionPolicy Bypass -File gsd-installer.ps1 --target "C:\path\to\repo" --mode missing`
-- `powershell -ExecutionPolicy Bypass -File gsd-installer.ps1 --target "C:\path\to\repo" --mode overwrite --backup`
-- `sh gsd-installer.sh --target "/path/to/repo" --mode missing`
-- `sh gsd-installer.sh --target "/path/to/repo" --mode overwrite --backup`
-
-### NPX installer (npm publish)
-
-After publishing this repo to npm as `its-magic`, install with:
-
-- `npx its-magic --target "<path>" --mode missing`
-
-Examples:
-
-- `npx its-magic --target "C:\path\to\repo" --mode missing`
-- `npx its-magic --target "C:\path\to\repo" --mode overwrite --backup`
-
-### NPX release automation
-
-1. `npm login`
-2. `npm publish`
-
-Or use:
-
-- `npm run release:patch`
-- `npm run release:minor`
-- `npm run release:major`
-
-### Unified release (npm + Chocolatey + Homebrew)
-
-Release to all three package managers with a single command:
-
-- Windows: `powershell -ExecutionPolicy Bypass -File scripts/release-all.ps1`
-- macOS/Linux: `sh scripts/release-all.sh`
-
-Or via npm scripts:
-
-**All three at once:**
-
-- `npm run release:all` — patch release to all three
-- `npm run release:all:patch` — explicit patch to all three
-- `npm run release:all:minor` — minor release to all three
-- `npm run release:all:major` — major release to all three
-- `npm run release:all:beta` — beta prerelease to all three
-- `npm run release:all:dry` — dry run (print actions, change nothing)
-
-**Single target only:**
-
-- `npm run release:npm-only` — publish only to npm
-- `npm run release:choco-only` — publish only to Chocolatey
-- `npm run release:brew-only` — publish only to Homebrew
-
-**Mix and match (PowerShell):**
-
-```powershell
-# All three, minor bump
-scripts/release-all.ps1 -Bump minor
-
-# Only npm + Homebrew (skip Chocolatey)
-scripts/release-all.ps1 -SkipChoco
-
-# Only npm + Chocolatey (skip Homebrew)
-scripts/release-all.ps1 -SkipBrew
-
-# Only npm
-scripts/release-all.ps1 -SkipChoco -SkipBrew
-
-# Only Chocolatey
-scripts/release-all.ps1 -SkipNpm -SkipBrew
-
-# Only Homebrew
-scripts/release-all.ps1 -SkipNpm -SkipChoco
-
-# Homebrew + automatic tap publish
-scripts/release-all.ps1 -SkipNpm -SkipChoco -BrewTapRepo USER/homebrew-tap
-
-# Homebrew update only (no tap push)
-scripts/release-all.ps1 -SkipNpm -SkipChoco -SkipBrewPush
-
-# Beta prerelease, only npm
-scripts/release-all.ps1 -Bump prerelease -NpmTag beta -SkipChoco -SkipBrew
-
-# Explicit version
-scripts/release-all.ps1 -Bump 1.2.3
-
-# Preview without changing anything
-scripts/release-all.ps1 -DryRun
-```
-
-**Mix and match (Bash):**
-
-```bash
-# All three, minor bump
-sh scripts/release-all.sh --bump minor
-
-# Only npm + Homebrew (skip Chocolatey)
-sh scripts/release-all.sh --skip-choco
-
-# Only npm
-sh scripts/release-all.sh --skip-choco --skip-brew
-
-# Only Chocolatey
-sh scripts/release-all.sh --skip-npm --skip-brew
-
-# Only Homebrew
-sh scripts/release-all.sh --skip-npm --skip-choco
-
-# Beta prerelease, only npm
-sh scripts/release-all.sh --bump prerelease --npm-tag beta --skip-choco --skip-brew
-
-# Preview without changing anything
-sh scripts/release-all.sh --dry-run
-```
-
-What the script does:
-1. Bumps version in `package.json`
-2. Publishes to npm (with optional dist-tag)
-3. Creates a GitHub release via `gh` CLI (marked as prerelease when version contains `-`)
-4. Updates Chocolatey `.nuspec`, computes checksum, runs `choco push`
-5. Updates Homebrew formula URL + sha256 (auto-selects `its-magic-beta.rb` for pre-releases)
-6. PowerShell script: pushes Homebrew formulas to your tap repo (`homebrew-tap`) unless `-SkipBrewPush` is used
-
-Prerequisites:
-- `npm login` (for npm)
-- `gh auth login` (for GitHub releases, optional but recommended)
-- `choco` installed + API key set (for Chocolatey, optional)
-- Homebrew tap repo for distributing the formula (`USER/homebrew-tap`)
-- `git` available locally (for tap push)
-
-### Installing via each package manager
-
-| Manager    | Stable                                    | Beta / Pre-release                          |
-|------------|-------------------------------------------|---------------------------------------------|
-| npm/npx    | `npx its-magic --target . --mode missing` | `npx its-magic@beta --target . --mode missing` |
-| Chocolatey | `choco install its-magic`                 | `choco install its-magic --pre`             |
-| Homebrew   | `brew install USER/tap/its-magic`         | `brew install USER/tap/its-magic-beta`      |
-
-### Homebrew tap setup
-
-To distribute via Homebrew you need a "tap" repo:
-
-1. Create a GitHub repo named `homebrew-tap` (e.g. `USER/homebrew-tap`)
-2. Copy both formulas into that repo:
-   - `packaging/homebrew/its-magic.rb` (stable)
-   - `packaging/homebrew/its-magic-beta.rb` (pre-releases)
-3. Users install with:
-   - Stable: `brew tap USER/tap && brew install its-magic`
-   - Beta: `brew tap USER/tap && brew install its-magic-beta`
-
-The PowerShell release script auto-detects whether the version is a pre-release
-(contains `-`, e.g. `0.2.0-beta.1`) and updates the correct formula, then
-commits + pushes to your tap repo.
-
-Tap push options (PowerShell):
-- `-BrewTapRepo USER/homebrew-tap` - explicit tap target
-- `-BrewTapBranch main` - target branch
-- `-BrewTapDir <path>` - local checkout path for tap
-- `-CreateBrewTapIfMissing $true|$false` - auto-create tap repo on GitHub (default: `$true`, requires `gh auth login`)
-- `-SkipBrewPush` - update formula locally but do not push tap
-
-### Chocolatey setup
-
-1. Create an account at https://community.chocolatey.org
-2. Get an API key from your account page
-3. Run `choco apikey --key <your-key> --source https://push.chocolatey.org/`
-4. The release script handles `choco pack` + `choco push` automatically
-
-```mermaid
-flowchart LR
-  ReleaseAll[scripts/release-all.*] --> VerCheck{version has -?}
-  VerCheck -->|stable| NPM["npm publish --tag latest"]
-  VerCheck -->|prerelease| NPMBeta["npm publish --tag beta"]
-  ReleaseAll --> GH["gh release create"]
-  VerCheck -->|prerelease| GHPre["--prerelease flag"]
-  GH --> Choco[choco pack + push]
-  GH --> BrewCheck{prerelease?}
-  BrewCheck -->|no| BrewStable[its-magic.rb]
-  BrewCheck -->|yes| BrewBeta[its-magic-beta.rb]
-  NPM --> U1["npx its-magic"]
-  NPMBeta --> U1b["npx its-magic@beta"]
-  Choco --> U2["choco install its-magic"]
-  Choco --> U2b["choco install its-magic --pre"]
-  BrewStable --> U3["brew install its-magic"]
-  BrewBeta --> U3b["brew install its-magic-beta"]
-```
-
-## Release package contents
-
-The npm package intentionally excludes **dev/testing/benchmark** content. The
-published package includes only the its-magic runtime (commands, rules, agents,
-docs, installers).
-
-Excluded from npm release:
-- `benchmarks/`
-- `tests/`
-- `packaging/`
-- `Plan.md`
-
-If you want those, use the development repo directly.
-
-## Benchmarks
-
-Use the benchmark harness to compare its-magic changes over time.
-
-- Windows: `powershell -ExecutionPolicy Bypass -File benchmarks/run-bench.ps1`
-- macOS/Linux: `sh benchmarks/run-bench.sh`
-
-Reports are written to `benchmarks/bench-report.md`.
-
-### Benchmark run diagrams
-
-```mermaid
-flowchart TD
-  StartBench[Start benchmark] --> SelectScenario[Load scenarios]
-  SelectScenario --> InstallKit[Install its-magic into temp workspace]
-  InstallKit --> RunChecks[Validate required files/sections]
-  RunChecks --> BenchReport[Write benchmarks/bench-report.md]
-```
-
-## Live benchmarks
-
-Live benchmarks capture real agent runs in Cursor via hook telemetry.
-
-- Windows: `powershell -ExecutionPolicy Bypass -File benchmarks/live/run-live-bench.ps1`
-- macOS/Linux: `sh benchmarks/live/run-live-bench.sh`
-
-Reports are written to `benchmarks/live/live-bench-report.md`.
-
-```mermaid
-flowchart TD
-  StartLive[Start live benchmark] --> SetSession[Set GSD_BENCH_SESSION]
-  SetSession --> RunCommands[Run /gsd-* commands in Cursor]
-  RunCommands --> LogHooks[Hook telemetry to gsd-bench-log.jsonl]
-  LogHooks --> LiveReport[Write live-bench-report.md]
-```
-
-## Prompted benchmark runs
-
-Use the prompt runner to step through a scenario like a human would by
-submitting each `/gsd-*` command in order.
-
-- Windows:
-  `powershell -ExecutionPolicy Bypass -File benchmarks/prompts/run-prompts.ps1 -PromptFile benchmarks/prompts/S4_webview_api_app.txt -Clipboard`
-- macOS/Linux:
-  `sh benchmarks/prompts/run-prompts.sh benchmarks/prompts/S4_webview_api_app.txt --clipboard`
-
-## Fully automated benchmark runs (Headless CLI)
-
-This uses Cursor CLI in non-interactive mode.
-
-- Windows: `powershell -ExecutionPolicy Bypass -File benchmarks/headless/run-headless.ps1 -PromptFile benchmarks/prompts/S4_webview_api_app.txt -ScenarioFile benchmarks/scenarios/S4_webview_api_app.scn`
-- macOS/Linux: `sh benchmarks/headless/run-headless.sh benchmarks/prompts/S4_webview_api_app.txt benchmarks/scenarios/S4_webview_api_app.scn`
-
-Dependencies:
-- Cursor CLI (`agent`)
-- ripgrep (`rg`)
-
-Outputs:
-- `benchmarks/headless/headless-report.md`
-- `benchmarks/headless/protocol.md` (step prompts + AI response summaries)
-Per-run artifacts:
-- `benchmarks/runs/headless-<timestamp>/workspace/`
-- `benchmarks/runs/headless-<timestamp>/reports/`
-
-```mermaid
-flowchart TD
-  StartHeadless[Start headless run] --> LoadPrompt[Load prompt blocks]
-  LoadPrompt --> TempWorkspace[Create temp workspace]
-  TempWorkspace --> InstallHeadless[Install its-magic]
-  InstallHeadless --> RunAgent[agent -p --force for each step]
-  RunAgent --> ValidateOutputs[Validate files/sections/smoke checks]
-  ValidateOutputs --> WriteProtocol[Write protocol.md]
-  WriteProtocol --> WriteHeadlessReport[Write headless-report.md]
-```
-
-## Autonomous dependency installs
-
-To enable fully autonomous dependency and runtime installs, set in
-`.cursor/scratchpad.md`:
-
-- `AUTO_INSTALL_DEPS=1`
-
-When enabled, the agent may install required dependencies using the relevant
-package manager without asking.
-
-## Remote / Docker execution
-
-To allow the agent to use remote or Docker servers for builds/tests, set:
-
-- `REMOTE_EXECUTION=1` in `.cursor/scratchpad.md`
-- Configure servers in `.cursor/remote.json`
-
-This only enables usage; you still control which servers are listed.
-
-## Release notes automation
-
-Enable automatic release notes:
-
-- Set `AUTO_RELEASE_NOTES=1` in `.cursor/scratchpad.md`
-- Run the generator (optional manual step):
-  - Windows: `powershell -ExecutionPolicy Bypass -File scripts/generate-release-notes.ps1`
-  - macOS/Linux: `sh scripts/generate-release-notes.sh`
-
-The generator writes to `handoffs/release_notes.md` and includes:
-- Sprint summary
-- Last 20 git commit messages (if in a git repo)
-- QA findings
-- Runbook notes
-
-## Automation modes
-
-Configure in `.cursor/scratchpad.md`:
-
-- `AUTO_FLOW_MODE=auto_until_decision` to run phases until a decision gate.
-- `PHASE_MODE=interactive|auto` to control per-phase prompting.
-- `PERMISSION_MODE=interactive|auto` to reduce routine permission prompts.
-
-### Scratchpad reference
-
-- `GSD_CONTEXT_STRICT=0|1` — enforce context refresh after code edits.
-- `LOOP_UNTIL_GREEN=0|1` — optional test loop when commands are set.
-- `RUN_TESTS_ON_EDIT=0|1` — run tests after edits.
-- `DONE=0|1` — stop hook loops.
-- `GSD_BENCH_SESSION=<id>` — live benchmark session id.
-- `AUTO_FLOW_MODE=manual|auto_until_decision`
-- `PHASE_MODE=interactive|auto`
-- `PERMISSION_MODE=interactive|auto`
-- `AUTO_INSTALL_DEPS=0|1`
-- `AUTO_RELEASE_NOTES=0|1`
-- `REMOTE_EXECUTION=0|1`
-- `REMOTE_CONFIG=.cursor/remote.json`
-
-## Recent changes (latest additions)
-
-- **Unified multi-platform release** (npm + Chocolatey + Homebrew) via `scripts/release-all.*`.
-- Homebrew formula template in `packaging/homebrew/`.
-- Chocolatey package spec in `packaging/chocolatey/`.
-- npm `release:all` scripts for one-command publishing to all three.
-- Headless CLI benchmark runner with protocol output.
-- Headless run validation: required files, sections, and smoke checks.
-- Prompted scenario files for code benchmarks.
-- Autonomous dependency installs (`AUTO_INSTALL_DEPS=1`).
-- Automated release notes (`AUTO_RELEASE_NOTES=1`).
-- Optional remote/Docker execution (`REMOTE_EXECUTION=1` + `remote.json`).
-
-```mermaid
-flowchart TD
-  Scratchpad[".cursor/scratchpad.md"] --> AutoDeps[AUTO_INSTALL_DEPS]
-  Scratchpad --> AutoRelease[AUTO_RELEASE_NOTES]
-  Scratchpad --> RemoteExec[REMOTE_EXECUTION]
-  AutoDeps --> ExecuteCmd[/gsd-execute/]
-  AutoRelease --> ReleaseCmd[/gsd-release/]
-  RemoteExec --> ExecuteCmd
-  ReleaseCmd --> ReleaseNotes[handoffs/release_notes.md]
-  BenchPrompts[benchmarks/prompts/*.txt] --> HeadlessRunner[benchmarks/headless/run-headless.*]
-  HeadlessRunner --> Protocol[benchmarks/headless/protocol.md]
-  HeadlessRunner --> Report[benchmarks/headless/headless-report.md]
-  ReleaseAll[scripts/release-all.*] --> NPMPub[npm publish]
-  ReleaseAll --> GHRelease[GitHub release]
-  GHRelease --> ChocoUpdate[Chocolatey pack+push]
-  GHRelease --> BrewUpdate[Homebrew formula update]
-```
-
-## How-to examples
-
-### Example 1: New web app idea
-
-1. Run `/gsd-intake` and describe the idea in one sentence.
-2. Answer the PO questions until `docs/product/*` are filled.
-3. Run `/gsd-research` to capture risks and patterns.
-4. Run `/gsd-architecture` and `/gsd-sprint-plan` to create `sprints/S0001/*`.
-5. Run `/gsd-plan-verify` to confirm task coverage.
-6. Run `/gsd-execute` to implement tasks and update `state.md`.
-7. Run `/gsd-qa` and record findings.
-8. Run `/gsd-verify-work` for UAT.
-9. Run `/gsd-release` and update `runbook.md`.
-10. Run `/gsd-refresh-context` to compact context.
-
-### Example 2: Pause and resume
-
-1. Run `/gsd-pause` to write `handoffs/resume_brief.md`.
-2. Later, run `/gsd-resume` to load the context pack.
-3. Continue with the appropriate phase.
-
-### Example 3: Map an existing project
-
-1. Run `/gsd-map-codebase` before any planning.
-2. Review `docs/engineering/codebase-map.md` and `dependencies.json`.
-3. Continue with `/gsd-intake` or `/gsd-architecture`.
-
-### Example 4: Headless benchmark run
-
-1. Install Cursor CLI and `rg` (ripgrep).
-2. Run:
-   `powershell -ExecutionPolicy Bypass -File benchmarks/headless/run-headless.ps1 -PromptFile benchmarks/prompts/S4_webview_api_app.txt -ScenarioFile benchmarks/scenarios/S4_webview_api_app.scn`
-3. Read:
-   `benchmarks/headless/headless-report.md` and `benchmarks/headless/protocol.md`.
-
-## Features (full list)
-
-- **Structured GSD workflow** with explicit phases and artifacts.
-- **Artifact-first memory**: everything important is persisted in files.
-- **Decision gate**: escalation to `decisions/DEC-xxxx.md`.
-- **Pause/resume**: checkpoint and resume brief support.
-- **Context refresh**: compact `state.md` and decisions.
-- **Voice input** supported as an input layer.
-- **CI/CD templates** driven by runbook commands.
-- **Hooks** for safety and context hygiene.
-- **Benchmarks** for repeatable validation and telemetry.
-- **Headless CLI** support for automation.
-- **Research phase** and plan verification loop.
-- **Phase context** files for preferences and constraints.
-- **Quick mode** for small tasks.
-- **Milestones** for larger multi-phase work.
-- **Map codebase** for existing projects.
-- **Remote/Docker execution** (optional).
-
-## Commands (what they do)
-
-- `/gsd-intake`: capture idea, ask questions, write vision/backlog/acceptance.
-- `/gsd-discovery`: collect design/UX references, update vision/backlog.
-- `/gsd-architecture`: define architecture, risks, and decisions.
-- `/gsd-sprint-plan`: create sprint + tasks + handoff to dev.
-- `/gsd-plan-verify`: verify sprint tasks cover acceptance criteria.
-- `/gsd-execute`: implement tasks; update summary/state.
-- `/gsd-qa`: test plan + findings + handoff back to dev.
-- `/gsd-verify-work`: user acceptance testing (UAT).
-- `/gsd-release`: release notes + runbook updates.
-- `/gsd-pause`: checkpoint + resume brief.
-- `/gsd-resume`: load context pack and continue.
-- `/gsd-refresh-context`: compact state + decisions + summary.
-- `/gsd-research`: research patterns, risks, and dependencies.
-- `/gsd-map-codebase`: map existing repos before changes.
-- `/gsd-quick`: short path for small tasks.
-- `/gsd-phase-context`: phase-specific preferences and constraints.
-- `/gsd-milestone-start`: initialize milestone tracking.
-- `/gsd-milestone-complete`: finalize milestone summary.
-- `/gsd-auto`: run phases until decision gate.
-
-## Full workflow example (end-to-end)
-
-This example shows when each command is best used and what it produces.
-
-1. **/gsd-intake**  
-   Use when you have only an idea.  
-   Output: `docs/product/vision.md`, `docs/product/backlog.md`,
-   `docs/product/acceptance.md`, `handoffs/po_to_tl.md`.
-
-2. **/gsd-discovery**  
-   Use when you need UX references or scope refinements.  
-   Output: updates to `vision.md` + `backlog.md`, refreshes `po_to_tl.md`.
-
-3. **/gsd-research**  
-   Use when you need patterns, risks, and dependency notes.  
-   Output: `docs/engineering/research.md`, `decisions.md`, `state.md`.
-
-4. **/gsd-architecture**  
-   Use when you need a concrete technical approach and risk notes.  
-   Output: `docs/engineering/architecture.md`, `decisions.md`, `state.md`.
-
-5. **/gsd-sprint-plan**  
-   Use when architecture is stable and you want a task list.  
-   Output: `sprints/S0001/sprint.md`, `tasks.md`, `progress.md`,
-   `handoffs/tl_to_dev.md`.
-
-6. **/gsd-plan-verify**  
-   Use to confirm tasks cover acceptance criteria.  
-   Output: `sprints/S0001/plan-verify.json`, `state.md`.
-
-7. **/gsd-execute**  
-   Use to implement tasks.  
-   Output: code changes, `sprints/S0001/summary.md`, `state.md`,
-   `handoffs/dev_to_qa.md` (when ready).
-
-8. **/gsd-qa**  
-   Use to verify and document findings.  
-   Output: `sprints/S0001/qa-findings.md`, `handoffs/qa_to_dev.md`,
-   updates to `state.md`.
-
-9. **/gsd-verify-work**  
-   Use for user acceptance testing (UAT).  
-   Output: `sprints/S0001/uat.json`, `uat.md`, `state.md`.
-
-10. **/gsd-release**  
-   Use to prepare release notes and finalize runbook commands.  
-   Output: `handoffs/release_notes.md`, updated `runbook.md`, updated `state.md`.
-
-11. **/gsd-refresh-context**  
-   Use after major changes to keep context compact.  
-   Output: compacted `state.md`, `decisions.md`, updated sprint summary.
-
-12. **/gsd-pause**  
-   Use any time you need to stop safely.  
-   Output: `handoffs/resume_brief.md`, updated `progress.md` and `state.md`.
-
-13. **/gsd-resume**  
-    Use to restart work from artifacts only.  
-    Output: updated `state.md` if needed, and a clear next step.
-
-## Full flow diagram
-
-```mermaid
-flowchart TD
-  Intake[/gsd-intake/] --> Discovery[/gsd-discovery/]
-  Discovery --> Research[/gsd-research/]
-  Research --> Architecture[/gsd-architecture/]
-  Architecture --> SprintPlan[/gsd-sprint-plan/]
-  SprintPlan --> PlanVerify[/gsd-plan-verify/]
-  PlanVerify --> Execute[/gsd-execute/]
-  Execute --> QA[/gsd-qa/]
-  QA -->|fixes needed| Execute
-  QA --> VerifyWork[/gsd-verify-work/]
-  VerifyWork --> Release[/gsd-release/]
-  Release --> Refresh[/gsd-refresh-context/]
-  Execute --> Pause[/gsd-pause/]
-  QA --> Pause
-  Release --> Pause
-  Pause --> Resume[/gsd-resume/]
-  Resume --> Execute
-```
-
-```mermaid
-flowchart TD
-  Start[Idea] --> Intake2[/gsd-intake/]
-  Intake2 -->|DecisionGate| Decision{DecisionRequired}
-  Decision -->|ChooseOption| Discovery2[/gsd-discovery/]
-  Decision -->|ChooseOption| Research2[/gsd-research/]
-  Decision -->|ChooseOption| Architecture2[/gsd-architecture/]
-  Decision -->|ChooseOption| SprintPlan2[/gsd-sprint-plan/]
-  Decision -->|ChooseOption| PlanVerify2[/gsd-plan-verify/]
-  Decision -->|ChooseOption| Execute2[/gsd-execute/]
-  Decision -->|ChooseOption| QA2[/gsd-qa/]
-  Decision -->|ChooseOption| VerifyWork2[/gsd-verify-work/]
-  Decision -->|ChooseOption| Release2[/gsd-release/]
-  Decision -->|ChooseOption| Refresh2[/gsd-refresh-context/]
-  Decision -->|ChooseOption| Pause2[/gsd-pause/]
-  Discovery2 --> Research2
-  Research2 --> Architecture2
-  Architecture2 --> SprintPlan2
-  SprintPlan2 --> PlanVerify2
-  PlanVerify2 --> Execute2
-  Execute2 --> QA2
-  QA2 -->|fixes needed| Execute2
-  QA2 --> VerifyWork2
-  VerifyWork2 --> Release2
-  Release2 --> Refresh2
-  Execute2 --> Pause2
-  QA2 --> Pause2
-  Release2 --> Pause2
-  Pause2 --> Resume2[/gsd-resume/]
-  Resume2 --> Execute2
-  Execute2 --> HandoffDevQA[handoffs/dev_to_qa.md]
-  QA2 --> HandoffQAD[handoffs/qa_to_dev.md]
-  Intake2 --> HandoffPOTL[handoffs/po_to_tl.md]
-  SprintPlan2 --> HandoffTLDev[handoffs/tl_to_dev.md]
-```
-
-## Rules (what they enforce)
-
-- `gsd-core.mdc`: phase flow, context pack, pause/resume, remote usage.
-- `gsd-quality.mdc`: small steps, tests/quality, optional auto-install.
-- `gsd-handoffs.mdc`: handoffs + state updates required.
-- `gsd-escalation.mdc`: decision gate and stop conditions.
-
-## Hooks (what they do)
-
-- `beforeShellExecution`: block dangerous commands.
-- `beforeReadFile`: warn on secret-like files.
-- `afterFileEdit`: track code edits vs context refresh.
-- `stop`: remind to refresh context when needed.
-
-## Artifacts (single source of truth)
-
-- `docs/product/*`: vision, backlog, acceptance.
-- `docs/engineering/*`: architecture, decisions, state, runbook.
-- `sprints/Sxxxx/*`: sprint scope, tasks, progress, QA findings, summary.
-- `decisions/*`: decision records.
-- `handoffs/*`: role-to-role transfer notes.
+# its-magic — AI dev team
+
+[GitHub Repository](https://github.com/fl0wm0ti0n/its-magic)
+
+You bring the idea; its-magic is your structured **AI dev team** in Cursor — PO, Tech
+Lead, Dev, QA, Release, and Curator — that turns ideas into shipped software through
+explicit phases and handoff artifacts.
+
+State lives in repo files (`docs/product`, `handoffs`, `sprints`, `decisions`) — not
+chat-only memory. Run `/intake` with your idea, then follow intake → discovery →
+architecture → sprint plan → execute → QA → release; pause/resume and decision gates
+keep you in control when you want to steer. Implementers: see `docs/developer/README.md`
+for the DEV shard.
+
+When you want hands-off delivery, enable **`AUTO_FLOW_MODE=full_autonomy`**
+(default-off), run **`/auto` once in Cursor**, and let the native in-chat auto-chain
+drain your backlog — self-verify UAT, bounded block retry, and advance to the next OPEN
+story or bug without re-invoking each phase manually. The outer driver is **optional**
+(**fallback** for headless/CI or when native chain is unavailable). Guided and
+decision-gated modes remain the default.
+
+## Features (what its-magic can do)
+
+### Autonomous AI workflow
+
+- Run `/intake` through `/release` with explicit phase handoffs and fresh subagent contexts.
+- Use `/pause` and `/resume` with checkpoints when you want to steer; escalate blocking
+  choices to `decisions/DEC-xxxx.md`.
+- Enable **`AUTO_FLOW_MODE=full_autonomy`** (default-off), run **`/auto` once in Cursor**,
+  and drain backlog in-chat; outer driver is **optional** / **fallback** for headless/CI.
+- Team mode routes work across PO, Tech Lead, Dev, QA, Release, and Curator roles.
+- Backlog and bug drain advance OPEN items without re-invoking each phase manually.
+- See the catalog in **Commands and workflow** for phase commands and orchestration details.
+
+### Quality & verification gates
+
+- 3-layer quality chain: AI execute/QA loop → local `validate-and-push` → CI auto-fix.
+- Phase gates include `/plan-verify`, `/qa`, `/verify-work`, and `/uat` with fail-closed stops.
+- `/acceptance` blocks README ↔ backlog drift; user-visible metadata guard on operator scripts.
+- Browser UAT probes with structured fallback when live browser checks are unavailable.
+- Release gates enforce coverage, parity, and evidence before publish.
+- See the catalog in **Features** (`/acceptance`) and **Commands and workflow** for gate commands.
+
+### Distribution & install
+
+- Global install via npm, npx, Chocolatey, or Homebrew; apply to any repo with
+  `its-magic --target`.
+- Modes: `missing` (safe merge), `overwrite` (+ `--backup`), `upgrade` (framework only),
+  and `--clean-repo`.
+- Lifecycle QA matrix validates fresh install, upgrade, backup, and clean-repo paths.
+- Multi-target release publish with confirmation gates for npm/choco/brew.
+- See the **Feature coverage catalog** below for distribution-tagged items.
+
+### Operator control & ergonomics
+
+- Scratchpad flags and `scratchpad.local.md` tune behavior without rewriting framework files.
+- Guided intake packs structure your first `/intake` conversation.
+- Caveman voice mode and optional input compression for terse operator UX.
+- `TOKEN_PROFILE` cost profiles slim context packs without changing workflow semantics.
+- Voice input shortcuts and permissions/runtime connectivity for remote execution.
+- See the catalog in **Other useful capabilities** for scratchpad and governance flags.
+
+<!-- readme-feature-coverage-catalog -->
+
+### Feature coverage catalog (US-0091)
+
+- `/acceptance` — README ↔ backlog/acceptance feature coverage backfill + blocking drift gate (`US-0091`).
+- `README.md` — Visionary intro + tiered feature hierarchy (autonomous AI dev team positioning, root/template parity) (`US-0094`).
+- `/auto` — Native in-chat auto-chain + full-autonomy mode (`US-0095`, `US-0092`).
+- `/bin` — POSIX npm installer + Linux remote test targets (WSL / SSH / Docker) (`US-0084`).
+- `/choco` — Configurable Multi-Target Release Publish with Confirmation Gate (`US-0054`).
+- `/devops` — First-Class Bug Issue Workflow (Open/Closed) (`US-0079`).
+- `/engineering` — Agent-Driven Codebase Map Bootstrap (`US-0082`).
+- `/engineering` — Remote Runtime Connectivity Contract for QA/Release/Publish (`US-0064`).
+- `/install` — Template/install payload omits intake gate scripts (`BUG-0001`).
+- `/installed` — its-magic ships its OWN packaging CI into generated repos, breaking CI in every created project (`BUG-0009`).
+- `/intake` — Optional Caveman-style input compression (safe file scope) (`US-0090`).
+- `/lint` — CI/CD Workflows (`US-0007`).
+- `/or` — Cursor Caveman mode (scratchpad-configurable terse responses) (`US-0089`).
+- `/push` — Multiplatform Distribution (`US-0009`).
+- `/run-tests` — Baseline Regression Cleanup for Installer and Version Sync Checks (`US-0074`).
+- `/template` — End-to-End Lifecycle QA for `its-magic` Install/Upgrade/Clean (`US-0041`).
+- `/upgrade` — Missing scripts still occur on install modes missing/upgrade (`BUG-0003`).
+- `/usr` — Global Linux install fails: empty `install_include_paths` when manifest is CRLF (`BUG-0008`).
+- `/workdir` — installer.sh fails in shell path with `set: Illegal option -` (`BUG-0004`).
+- `MIGRATION` scratchpad flag — Smart Upgrade Mode (`US-0018`).
+- `US-0016` scratchpad flag — Homebrew Version Sync (`US-0016`).
+
+## Setup
+
+its-magic is an installer you run once per repo. It copies the AI dev team
+workflow files (`.cursor/` commands, rules, agents, hooks, skills, plus `docs/`,
+`sprints/`, `handoffs/`, etc.) into your project.
+
+Starter artifacts are shipped as clean placeholders (no preloaded sprint/demo
+history), so `/intake` starts from your own idea.
+
+### 1) Install its-magic (once)
+
+Pick one method:
+
+| Method | Install command |
+|--------|----------------|
+| npm    | `npm install -g its-magic` |
+| npx    | `npx its-magic --target . --mode missing` |
+| Chocolatey | `choco install its-magic` (Admin shell) |
+| Homebrew | `brew tap USER/tap && brew install its-magic` |
+
+### Global Linux install: empty `install_include_paths` (CRLF manifest)
+
+If **`its-magic --target <repo> --mode missing`** fails with **`[INSTALL_MANIFEST_ERROR] install_include_paths section is empty`** on Debian/Linux while the packaged manifest still lists paths, the global install likely has **CRLF** line endings in **`installer-owned-paths.manifest`** (visible as **`^M$`** with **`cat -A`**). **Fix in-tree** from **`0.1.2-41`**: **`installer.sh`** strips trailing carriage returns before section matching; **`.gitattributes`** keeps **`*.manifest`** LF; **`prepublishOnly`** runs **`guard_installer_publish`**. **Upgrade**: install a build **≥ `0.1.2-41`** (or reinstall from a fresh **`npm pack`** tarball after pull). Older tarballs such as **`its-magic@0.1.2-40`** may remain broken until republished — see **`docs/engineering/architecture.md`** **`# BUG-0008`**.
+
+### 2) Apply to a repo
+
+New repo:
+
+```bash
+mkdir my-project && cd my-project
+git init
+its-magic --target . --mode missing --create
+```
+
+Existing repo (safe merge):
+
+```bash
+its-magic --target . --mode missing
+```
+
+Existing repo (overwrite + backup):
+
+```bash
+its-magic --target . --mode overwrite --backup
+```
+
+### Upgrading an existing repo
+
+When you update its-magic to a newer version (`npm update -g its-magic`), run
+upgrade mode to update framework files while preserving your project data:
+
+```bash
+its-magic --target . --mode upgrade
+```
+
+What upgrade does:
+
+- **Framework files** (commands, rules, agents, hooks, skills, CI, scripts) are
+  updated to the latest version.
+- **User data** (docs, sprints, handoffs, decisions, runbook) is never touched.
+- **Mixed files** (`README.md`) are preserved. If the template version has new
+  content, a review notice is printed.
+- **Scratchpad baseline (DEC-0055 / US-0073, Model B):** `.cursor/scratchpad.md`
+  is not copied as a manifest file; the installer **materializes** it from the
+  packaged template when missing and validates required merged keys (Python
+  required). Legacy repos that already committed `.cursor/scratchpad.md` keep it on
+  upgrade (not overwritten).
+- A canonical version marker is stored at `its_magic/.its-magic-version` in your repo.
+- Installer bootstrap is OS-aware + stack-aware for runbook command defaults
+  (`TEST_COMMAND`, optional `LINT_COMMAND`/`TYPECHECK_COMMAND`) and preserves
+  explicit user overrides.
+
+Upgrade with backup (backs up framework files before updating):
+
+```bash
+its-magic --target . --mode upgrade --backup
+```
+
+### 3) Open in Cursor
+
+1. Open the project folder
+2. Run `/intake` with your idea
+3. Follow the workflow
+
+### CLI quick commands
+
+```bash
+# Show banner + help
+its-magic
+
+# Show version only
+its-magic --version
+
+# Install workflow files into current repo
+its-magic --target . --mode missing
+
+# Clean previously installed workflow artifacts
+its-magic --clean-repo --target .
+```
+
+### Installer options
+
+**Install options**
+
+| Flag | Description |
+|------|-------------|
+| `--target <path>` | Path to the repository where workflow files are installed. If omitted you are prompted interactively. |
+| `--mode missing` | **Default.** Only copy files that do not exist yet. Safe for repos that already have some workflow files. |
+| `--mode overwrite` | Replace every file, even if it already exists. Combine with `--backup` to keep a snapshot first. |
+| `--mode interactive` | Ask per file whether to overwrite or skip. Useful when you want to cherry-pick updates. |
+| `--mode upgrade` | Update framework files (commands, rules, agents, hooks, skills, CI, scripts) while preserving user data (docs, sprints, handoffs, decisions). Use after updating its-magic to a newer version. |
+| `--backup` | Before overwriting, save existing files to `backups/<timestamp>/`. Ignored in `missing` mode (nothing gets replaced). |
+| `--create` | Create the target directory if it does not exist. |
+
+**Clean options**
+
+| Flag | Description |
+|------|-------------|
+| `--clean-repo` | Remove installer-owned its-magic workflow artifacts from the target repo (manifest-owned paths including `.cursor`, `docs/product`, `docs/engineering`, `docs/user-guides`, `sprints`, `handoffs`, `decisions`, workflow scripts, CI files, installer metadata in `its_magic/`, and legacy `.its-magic-version`). Your own source code is never touched. |
+| `--yes` | Skip the confirmation prompt when cleaning. |
+
+**Info**
+
+| Flag | Description |
+|------|-------------|
+| `--help`, `-h` | Show banner, version, repo URL, and full usage reference. |
+| `--version`, `-v` | Print the installed its-magic version and exit. |
+
+### Lifecycle QA matrix (US-0041)
+
+`its-magic` lifecycle behavior is validated in both installer and CLI paths.
+Primary coverage:
+
+| Scenario | Local coverage | CI coverage | Expected evidence |
+|---|---|---|---|
+| Fresh install (`missing`) | `tests/run-tests.ps1`, `tests/run-tests.sh` | npm/brew/choco jobs | Required files + `its_magic/.its-magic-version` |
+| Overwrite + backup | `tests/run-tests.ps1`, `tests/run-tests.sh` | lifecycle subset in CI jobs | Backup snapshot contains overwritten framework file |
+| Upgrade lifecycle | `tests/run-tests.ps1`, `tests/run-tests.sh`, npm local package tests | lifecycle subset in CI jobs | Framework file restored, user-data preserved |
+| Clean-repo safety | `tests/run-tests.ps1`, `tests/run-tests.sh`, npm local package tests | lifecycle subset in CI jobs | Framework artifacts removed, non-framework marker preserved |
+| Negative-path invalid mode/args | `tests/run-tests.ps1`, `tests/run-tests.sh` | lifecycle subset in CI jobs | Non-zero fail-fast behavior |
+
+Run locally:
+
+```bash
+sh tests/run-tests.sh
+powershell -ExecutionPolicy Bypass -File tests/run-tests.ps1
+```
+
+## How-to
+
+### Command usage pattern
+
+- Best practice: use `/<command>` + 1-3 lines context.
+- For quick ops (`/pause`, `/resume`, `/refresh-context`) command-only is fine.
+
+### What gets installed
+
+```text
+your-project/
+  .cursor/commands/          Cursor slash commands
+  .cursor/rules/             AI behavior rules
+  .cursor/agents/            Subagent definitions
+  .cursor/skills/            Reusable skills
+  .cursor/hooks/             Automation hooks
+  .cursor/scratchpad.md      Materialized shared defaults (Model B; not manifest-copied)
+  .cursor/scratchpad.local.example.md   Framework default key catalog
+  docs/                      Engineering & product docs, runbook
+  sprints/                   Sprint tracking artifacts
+  handoffs/                  Phase handoff artifacts
+  decisions/                 Decision records
+  scripts/validate-and-push.ps1   Local test-fix-push loop (Windows)
+  scripts/validate-and-push.sh    Local test-fix-push loop (Linux/Mac)
+  .github/workflows/         CI with auto-fix loop
+  README.md
+```
+
+### Team mode local overrides (recommended)
+
+Use three layers (merge precedence: **local > materialized baseline > example**,
+`DEC-0055`):
+
+- Framework catalog: `.cursor/scratchpad.local.example.md` (installed; refreshed on upgrade)
+- Shared team baseline: `.cursor/scratchpad.md` (materialized on install when missing; commit as you prefer)
+- Personal overrides: `.cursor/scratchpad.local.md` (gitignored; never overwritten by install/upgrade)
+
+Setup:
+
+1. Run `its-magic` — baseline is materialized and merged validation runs (requires Python on PATH for `installer.ps1` / `installer.sh`).
+2. Optionally copy `.cursor/scratchpad.local.example.md` to `.cursor/scratchpad.local.md` for personal values (`TEAM_MEMBER`, `ACTIVE_TASK_IDS`, …).
+
+Recovery if `.cursor/scratchpad.md` is missing or merge validation fails:
+
+```bash
+python installer.py --scratchpad-postinstall --target . --mode missing
+```
+
+Upgrade behavior (US-0057 / DEC-0057):
+- Aligns with **DEC-0039** (example vs local ownership), **DEC-0057** (example-first
+  ordering relative to baseline materialization), and Model B baseline rules below.
+
+- `.cursor/scratchpad.local.example.md` is framework-owned and always refreshed from
+  the shipped template during post-install **before** baseline handling (`DEC-0057` **AC-1..AC-3**).
+- `.cursor/scratchpad.local.md` is user-owned and preserved on `--mode upgrade`.
+- Existing `.cursor/scratchpad.md` is left untouched on upgrade unless missing (then
+  materialized) or `overwrite` / fresh materialize paths apply (Model B).
+- Installer output uses `[SCRATCHPAD_LAYER]` lines to distinguish example refresh,
+  baseline materialize/skip, and user-local preservation (`DEC-0057` **AC-5**).
+- Paired catalog parity (baseline vs `.cursor/scratchpad.local.example.md`, active and
+  `template/`): `python scripts/check-scratchpad-pair-parity.py --repo .` (wired into
+  `tests/run-tests.ps1` / `tests/run-tests.sh`; **AC-11**).
+
+Deterministic ordering behavior (US-0058):
+- Mutable artifacts follow `docs/engineering/artifact-ordering-policy.md`.
+- `state.md` checkpoints are append-bottom; `backlog.md` and `acceptance.md`
+  remain sorted-canonical by story ID.
+- Commands fail closed on ambiguous placement anchors using
+  `ARTIFACT_ORDERING_ANCHOR_AMBIGUOUS`.
+- Commands fail closed on non-monotonic state checkpoint timestamps using
+  `STATE_TIMESTAMP_NON_MONOTONIC`.
+
+Intake runtime safety behavior (US-0059):
+- `/intake` requires role-specific `po` capability by default and fails fast with
+  `SUBAGENT_CAPABILITY_UNAVAILABLE` when unavailable.
+- Silent in-band fallback is disabled by default and only allowed with explicit
+  `INTAKE_SUBAGENT_FALLBACK=allow`.
+- Drift detection distinguishes self-write updates from external concurrent
+  writers; true conflicting external writes fail safe with
+  `INTAKE_CONCURRENT_WRITER_DETECTED`.
+
+Runtime QA autopilot behavior (US-0065):
+- Generated-project QA must include runtime proof chain:
+  `startup -> readiness/connectivity -> log scan -> bounded retry -> verdict`.
+- Deterministic runtime fail codes:
+  `RUNTIME_STARTUP_FAILED`, `RUNTIME_ENDPOINT_UNREACHABLE`,
+  `RUNTIME_LOG_CRITICAL_DETECTED`, `RUNTIME_RETRY_BUDGET_EXHAUSTED`,
+  `RUNTIME_STACK_PROFILE_UNRESOLVED`.
+- Runtime evidence must include startup command/profile, runtime mode
+  (`local|remote`), health result, retry ledger, and log severity summary.
+- Stack-aware runtime profile resolution is required for Node/Python/Go/Java/.NET;
+  unresolved stacks fail closed (no generic silent PASS fallback).
+- For webapp contexts, QA includes browser-surface verification with
+  console/network error signals.
+
+Generated test scaffolding + auto-run behavior (US-0066):
+- `/execute` resolves stack profile (`node|python|go|java|dotnet`) and generates
+  missing baseline unit/integration/acceptance tests only.
+- Generation is non-destructive by default: preserve user-authored tests/config,
+  fill only missing baseline assets, keep reruns idempotent.
+- `TEST_COMMAND` wiring is deterministic:
+  - preserve existing non-empty user command,
+  - set stack baseline only when command is missing/unset.
+- `/qa` automatically runs the generated baseline tests and records deterministic
+  evidence (`command`, `result`, `output ref`, `generated paths ref`).
+- Fail-closed scaffold diagnostics:
+  `TEST_SCAFFOLD_STACK_UNRESOLVED`,
+  `TEST_SCAFFOLD_UNSUPPORTED_STACK`,
+  `TEST_SCAFFOLD_GENERATION_FAILED`.
+- Static baseline test pass does not bypass runtime autopilot; runtime verdict
+  remains mandatory for QA PASS.
+
+## Commands and workflow
+
+### Core commands
+
+- `/ask`: ask questions using project context (read-only, no artifacts created).
+- `/intake`: capture idea, backlog, acceptance.
+- `/discovery`: collect UX/product references.
+- `/research`: risks, patterns, dependencies.
+- `/architecture`: technical approach and decisions.
+- `/sprint-plan`: sprint and task list.
+- `/plan-verify`: acceptance coverage check.
+- `/execute`: implement tasks.
+- `/qa`: test and report findings.
+- `/verify-work`: UAT.
+- `/release`: release notes + runbook updates.
+- `/memory-audit`: read-only memory drift check with advisory report.
+- `/pause`, `/resume`, `/refresh-context`.
+- `/auto`: orchestration mode that spawns a fresh subagent per phase.
+
+### Guided intake behavior (US-0033)
+
+`/intake` supports two PO interaction modes via `.cursor/scratchpad.md`:
+
+- `INTAKE_GUIDED_MODE=1` (default)
+  - asks targeted follow-up only when needed for concrete acceptance
+  - presents options/alternatives before recommendation
+  - preserves user decision authority
+  - runs intake-time research and persists R-xxxx evidence
+- `INTAKE_GUIDED_MODE=0` (low-touch)
+  - skips proactive follow-up/options/research overhead unless user requests it
+  - still performs duplicate/overlap check against backlog
+
+### Intake decomposition + risk-aware questioning (US-0051)
+
+When guided mode is enabled, `/intake` now supports bounded decomposition for
+broad/high-risk requests:
+
+- runs deterministic breadth/risk heuristics (feature/workflow count,
+  cross-cutting impact, acceptance breadth, unknown dependencies)
+- proposes bounded multi-story decomposition when heuristics indicate broad
+  scope; keeps single-story default for narrow scope
+- enforces vertical-slice/workflow-step split quality (independently valuable,
+  testable stories; avoid technical-layer-only splits by default)
+- preserves user control before persistence: accept, merge, or adjust split
+- asks additional targeted questions on high-risk/high-impact intake (not
+  ambiguity-only), but keeps rounds bounded and concise
+- keeps low-touch compatibility: no forced decomposition when
+  `INTAKE_GUIDED_MODE=0` unless explicitly requested
+- records decomposition/questioning evidence in intake artifacts
+  (`docs/product/backlog.md`, `docs/product/acceptance.md`,
+  `handoffs/po_to_tl.md`)
+
+### Mandatory intake question packs (US-0068)
+
+`/intake` now enforces deterministic minimum questionnaire packs before
+backlog/acceptance persistence:
+
+- `first-intake-pack` for first/new/broad requests
+- `small-intake-pack` for narrow follow-up requests
+
+Fail-closed coverage behavior:
+
+- required topic answers must be covered for the selected pack before write
+- unknown/ambiguous stack cues fail closed to `first-intake-pack`
+- persistence blocks with deterministic reason codes when required coverage is
+  incomplete and assumptions are not explicitly confirmed
+
+Deterministic reason codes:
+
+- `INTAKE_REQUIRED_TOPIC_MISSING`
+- `INTAKE_REQUIRED_PACK_INCOMPLETE`
+- `INTAKE_ASSUMPTION_CONFIRMATION_REQUIRED`
+- `INTAKE_PERSISTENCE_BLOCKED`
+
+Intake artifacts must persist coverage evidence fields:
+
+- `asked_topics`
+- `missing_topics`
+- `assumptions_confirmed`
+
+### Interactive intake evidence + validator (US-0078 / DEC-0060)
+
+**US-0078** closes silent persistence: every intake that mutates backlog/acceptance must pass the
+deterministic **`intake_evidence`** gate — **`topic_coverage`** with valid **`ie:`** refs,
+asked-vs-covered alignment, and **`assumption_confirmation_ref`** when assumptions are affirmative.
+
+- Run `python scripts/intake_evidence_validate.py --self-test` (also exercised via `tests/run-tests.*` §26k).
+- **Packaged installs (BUG-0001 / DEC-0063)**: the intake gate modules (`intake_evidence_validate.py`, `intake_evidence_lib.py`, `intake_bug_routing_guard.py`) ship under **`template/scripts/`** and hydrate consumer repos at **`scripts/`** (npm **`files`**, Chocolatey/Homebrew **`template/`** tree, **`installer.ps1` / `installer.sh`** + **`installer-owned-paths.manifest`**). **`--mode upgrade`** treats them as framework files (added/updated like other shipped scripts). CI parity: **`python scripts/check_intake_template_parity.py --repo .`** (`tests/run-tests.*` §26N).
+- Operator docs: **`decisions/DEC-0060.md`**, **`docs/engineering/architecture.md`** **`# US-0078`**, runbook section **Interactive intake evidence validation (US-0078 / DEC-0060)**.
+- **Guided** and **low-touch** share the **same pre-persistence validation pipeline**; low-touch does not bypass mandatory pack coverage.
+
+### Bug issues + intake routing (US-0079 / DEC-0061)
+
+Defects use **`BUG-####`** under **`docs/product/backlog.md`** **`## Bug issues (canonical)`** with **`OPEN`/`DONE`** only and minimum reproducibility fields. Intake must not silently file defect prose as **`US-xxxx`**: set merged scratchpad **`INTAKE_WORK_ITEM_KIND=bug`** and/or use **`/intake bug`**, then run **`python scripts/intake_bug_routing_guard.py --kind story --file <prose.txt>`** before story allocation when in doubt.
+
+- Validators: `python scripts/bug_issue_validate.py --self-test`; `python scripts/bug_issue_validate.py --backlog docs/product/backlog.md --check-acceptance`.
+- Operator docs: **`decisions/DEC-0061.md`**, **`docs/engineering/architecture.md`** **`# US-0079`**, runbook **Bug issues (US-0079 / DEC-0061)**.
+
+### Optional ID namespace bootstrap (US-0052)
+
+Fresh-project ID bootstrap behavior is explicit and default-off:
+
+- `ID_NAMESPACE_BOOTSTRAP=0|1` in `.cursor/scratchpad.md` (default `0`)
+
+When enabled (`1`), workflows use deterministic freshness checks before first ID
+creation:
+
+- no `US-` IDs in `docs/product/backlog.md`
+- no `DEC-` IDs in `docs/engineering/decisions.md` / `decisions/DEC-*.md`
+- no `R-` IDs in `docs/engineering/research.md`
+
+If eligible, first IDs start at `US-0001`, `DEC-0001`, and `R-0001`. If not
+eligible (or mode is off), generation continues from highest existing IDs.
+Historical IDs are never rewritten or renumbered. Ineligible bootstrap requests
+emit deterministic diagnostic `ID_BOOTSTRAP_NOT_FRESH`.
+
+### Context compaction + tiered token profile (US-0053)
+
+Token-cost behavior is controlled by `.cursor/scratchpad.md`:
+
+- `TOKEN_PROFILE=lean|balanced|full` (default `balanced`)
+
+Profile behavior:
+
+- `lean`: reduce non-critical overhead defaults (automation/research/context
+  breadth) while keeping mandatory quality gates intact.
+- `balanced`: preserve current capabilities with moderate overhead.
+- `full`: maximize context breadth/autonomy for high-uncertainty work.
+
+Manual override precedence:
+
+- Explicit scratchpad flag values override profile defaults for that flag.
+- Profile mode never disables mandatory `/qa` -> `/verify-work` -> `/release`
+  gate semantics.
+
+Compaction behavior:
+
+- `docs/engineering/state.md` is the active hot surface.
+- Historical checkpoints move to append-only packs under
+  `docs/engineering/state-archive/`.
+- `docs/engineering/decisions.md` stays a compact index with bounded summaries
+  and canonical links to `decisions/DEC-xxxx.md`.
+- Enforced rollover thresholds:
+  - `STATE_HOT_MAX_LINES` (default `1200`)
+  - `STATE_HOT_MAX_CHECKPOINTS` (default `80`)
+  - `PO_TO_TL_HOT_MAX_LINES` (default `800`)
+  - `PO_TO_TL_HOT_MAX_SECTIONS` (default `60`)
+  - `ARCH_HOT_MAX_LINES` (default `3500`)
+  - `ARCH_HOT_MAX_STORY_SECTIONS` (default `120`)
+  Triad hot surfaces (`state.md`, `handoffs/po_to_tl.md`,
+  `docs/engineering/architecture.md`) must stay within merged scratchpad caps.
+
+### Token-cost measurement and low-cache patterns (US-0080 / DEC-0062)
+
+- Prefer **fresh subagent/chat boundaries** per `/auto` phase spawn (see `.cursor/commands/auto.md`).
+- Use explicit **`/auto start-from=<phase>`** when resuming so **`resolved_phase_plan`**
+  intersection stays deterministic (**`DEC-0052`**).
+- Select **`TOKEN_PROFILE=lean`** when compatible with your work to reduce scratchpad-driven
+  breadth; mandatory gates (**`US-0048`**, **`US-0056`**, **`US-0069`**, **`US-0039`**) stay on.
+- **Comparable** cache-read baselines require identical **`run_class_hash`**; otherwise
+  **`TOKEN_COST_RUN_CLASS_MISMATCH`** (no cross-plan gaming).
+- Committed metrics: **`handoffs/token_cost_runs/<orchestrator_run_id>.md`**; link from
+  **`docs/engineering/state.md`** via **`token_cost_evidence_ref`**.
+- Tooling: **`scripts/token_cost_lib.py`**, **`scripts/token_cost_compare.py`**,
+  **`python scripts/check_token_cost_parity.py --repo .`**.
+  Use `python scripts/enforce-triad-hot-surface.py --check` before completing a
+  phase that mutates them; use `--rollover` to archive oldest material into
+  deterministic packs when over cap (DEC-0054).
+  Archive verification mismatch fails with
+  `STATE_ARCHIVE_VERIFICATION_FAILED`.
+
+### Cross-phase artifact ownership guard (US-0061)
+
+To prevent accidental history loss across workflow phases:
+
+- canonical ownership policy: `docs/engineering/artifact-ownership-policy.md`
+- non-authorized phases must not delete or rewrite other-phase owned sections
+- `docs/engineering/architecture.md` is history-preserving (append or
+  target-section-only mutation)
+- deterministic fail-safe diagnostics:
+  `PHASE_OWNERSHIP_VIOLATION`,
+  `PHASE_OVERRIDE_EVIDENCE_MISSING`,
+  `ARCH_HISTORY_DELETION_DETECTED`
+
+`/ask` policy (read-only):
+
+- question-scoped retrieval first
+- targeted sections before broad file reads
+- bounded expansion only when unresolved
+- explicit "not found in artifacts" when still unresolved
+
+### Configurable multi-target publish + confirmation gate (US-0054)
+
+Post-release publish behavior is configurable per repository:
+
+- `RELEASE_PUBLISH_MODE=disabled|confirm|auto` (default `confirm`)
+- `RELEASE_TARGETS_FILE=docs/engineering/release-targets.json`
+- `RELEASE_TARGETS_DEFAULT=` optional comma-separated default targets
+
+Supported target types include:
+
+- `npm`, `choco`, `brew`, `git`, `docker`, `cloud`
+- `custom` (generic command target)
+- `ssh` (generic server deployment over SSH)
+- Connectivity metadata for remote/local operator context:
+  - `runtime.mode` (`local|remote`)
+  - endpoint fields (`domainEnv|ipEnv|hostEnv`, `port`, `protocol`)
+  - optional Traefik/ingress metadata
+  - optional `dockerOverSsh` contract for remote Docker execution over SSH
+
+Safety defaults:
+
+- Mandatory `/release` gates are unchanged and must pass first.
+- `confirm` mode enforces explicit operator approval before publish execution.
+- Sensitive values are env-referenced (for example `tokenEnv`, `authEnv`), not
+  inline literals.
+- Remote connectivity config errors fail fast with
+  `REMOTE_CONNECTIVITY_CONFIG_INVALID`.
+- Release/QA outputs use canonical operator connectivity doc:
+  `docs/engineering/runtime-connectivity.md`.
+
+### Deterministic status reconciliation command (US-0055)
+
+Use `/status-reconcile` to normalize status drift between canonical and derived
+workflow artifacts before continuation:
+
+- canonical source: `docs/product/backlog.md` story status
+- derived targets: `docs/product/acceptance.md`, `docs/engineering/state.md`,
+  `handoffs/resume_brief.md`
+- deterministic outcomes: apply/no-op/fail-safe reason codes with audit evidence
+  in `docs/engineering/status-normalization-report.md`
+
+This command is the bounded repair counterpart to `/memory-audit`
+(read-only detection).
+
+### Optional cross-repo observability (US-0034)
+
+Use optional compatibility visibility with default-safe off behavior:
+
+- `CROSS_REPO_OBSERVABILITY=0|1` (default `0`)
+- `COMPATIBILITY_GATE_ON_CRITICAL=0|1` (default `1`)
+- `COMPATIBILITY_SOURCES=` monitored `repo/module/contract/docs` declarations
+
+When disabled (`0`), workflow adds zero required compatibility overhead.
+
+When enabled (`1`), compatibility signals/findings are tracked in:
+
+- `docs/engineering/compatibility-signals.md`
+- `docs/engineering/compatibility-report.md`
+- `docs/engineering/manifests/registry.manifest.yaml`
+- `docs/engineering/manifests/repo.manifest.yaml`
+
+If unresolved critical findings remain and
+`COMPATIBILITY_GATE_ON_CRITICAL=1`, release progression must stop for a
+decision gate (`COMPATIBILITY_CRITICAL_OPEN`).
+
+### Optional component-scoped execution (US-0035)
+
+Enable scoped workflow behavior with:
+
+- `COMPONENT_SCOPE_MODE=0|1` (default `0`)
+- `TARGET_COMPONENTS=<comma-separated-component-ids>`
+
+When disabled (`0`), workflow adds zero required scope overhead.
+
+When enabled (`1`):
+
+- Scope declaration is tracked in `docs/engineering/component-scope.md`.
+- Sprint tasks should declare target components and expected impacted interfaces.
+- QA records unaffected-component protection checks in
+  `docs/engineering/component-scope-report.md`.
+- Unapproved out-of-scope impact must block release via decision gate
+  (`COMPONENT_SCOPE_VIOLATION_UNAPPROVED`).
+
+### Optional spec-pack documentation (US-0031)
+
+Optional Design Concept, CRS, and Technical Specification artifacts are
+controlled by:
+
+- `SPEC_PACK_MODE=0|1` (default `0`)
+
+When disabled (`0`), intake/architecture/execute/qa/release add no required
+spec-pack steps (zero overhead).
+
+When enabled (`1`):
+
+- Canonical paths per story: `docs/engineering/spec-pack/<story_id>-design-concept.md`,
+  `docs/engineering/spec-pack/<story_id>-crs.md`,
+  `docs/engineering/spec-pack/<story_id>-technical-specification.md`.
+- Minimum required sections and ownership are in `docs/engineering/runbook.md`.
+- Release gate validates completeness and blocks with `SPEC_PACK_INCOMPLETE` when
+  required sections are missing.
+
+### Optional user-guide documentation (US-0032)
+
+Optional per-feature user guides (end-user how-to docs) are controlled by:
+
+- `USER_GUIDE_MODE=0|1` (default `0`)
+
+When disabled (`0`), intake/architecture/sprint-plan/execute/qa/release add no
+required user-guide steps or blocking checks (zero overhead).
+
+When enabled (`1`):
+
+- Canonical path per feature story: `docs/user-guides/US-xxxx.md`.
+- Minimum required sections: Purpose, Prerequisites, Usage steps, Example,
+  Limitations, Troubleshooting (see `docs/engineering/runbook.md` and
+  `docs/user-guides/README.md`).
+- Release gate validates guide completeness and blocks with `USER_GUIDE_INCOMPLETE`
+  when enabled and required sections are missing.
+- User guides are end-user only; they do not duplicate spec-pack (US-0031) content.
+
+### Release notes model (US-0040)
+
+Release history is sprint-scoped and queue-backed:
+
+- Canonical sprint notes: `handoffs/releases/Sxxxx-release-notes.md`
+- Canonical queue tracker: `handoffs/release_queue.md`
+- Legacy compatibility pointer: `handoffs/release_notes.md`
+
+Deterministic release semantics:
+- Only target sprint artifacts/queue row may be mutated during one `/release` run.
+- Entering release flow sets target row to `unreleased`.
+- Successful finalization transitions same row to `released`.
+- Unresolved sprint identity or queue/notes mismatch fails closed with reason
+  codes and remediation guidance; no destructive reconciliation by default.
+
+### Post-QA release issue workflow (US-0042)
+
+Release gate chain (US-0039): `/release` enforces mandatory gates in order — check-in test, QA completion, UAT completion — then finalization. Blank optional runbook keys (`LINT_COMMAND`, `TYPECHECK_COMMAND`) do not block release; they are reported as skipped.
+
+If a problem appears **after QA** (during `/release`), record it separately from
+QA findings:
+
+- Release findings artifact: `sprints/Sxxxx/release-findings.md`
+- Release-to-dev handoff: `handoffs/release_to_dev.md`
+
+Boundary:
+- QA-phase issues -> `sprints/Sxxxx/qa-findings.md`
+- Post-QA release-gate issues -> `sprints/Sxxxx/release-findings.md`
+
+Each blocked release finding should include reason code, evidence refs,
+remediation, and rerun criteria.
+
+### Backlog reconciliation invariant (US-0043)
+
+Release completion must not leave stale backlog status for target sprint stories.
+At release finalization:
+
+- reconcile target story status to `DONE` using canonical release evidence;
+- reconcile target story acceptance checkboxes to checked state;
+- mutate only target sprint stories (never unrelated backlog entries);
+- fail safe with `BACKLOG_STATUS_DRIFT` if contradiction remains (e.g. released
+  sprint but backlog still `OPEN`/unchecked).
+
+### Canonical story status + normalization guard (US-0045)
+
+- `docs/product/backlog.md` is canonical for story `OPEN|DONE` status.
+- `docs/product/acceptance.md` and `docs/engineering/state.md` are derived views
+  reconciled from canonical backlog status plus release evidence.
+- One-time normalization baseline is recorded in
+  `docs/engineering/status-normalization-report.md`.
+- Contradictory resolution at release/reconciliation boundaries fails safe with:
+  - `BACKLOG_STATUS_DRIFT`
+  - `CANONICAL_STATUS_CONFLICT`
+
+### Agent isolation model
+
+- Every phase command runs in a fresh agent/subagent context.
+- Handoff files are the only cross-phase memory (`handoffs/*.md` + artifact
+  files).
+- Never rely on "ignore prior chat"; use a new context boundary instead.
+- `/auto` is orchestration only: it calls phase subagents and transfers context
+  through artifacts.
+
+#### Per-phase isolation evidence (US-0048 / DEC-0029)
+
+Isolation is enforced with auditable evidence written to `docs/engineering/state.md`.
+Each phase run appends:
+
+- `phase_id`, `role`, `fresh_context_marker`, `timestamp`, `evidence_ref`
+
+Missing/invalid/stale evidence fails closed with reason codes:
+`PHASE_CONTEXT_ISOLATION_MISSING`, `PHASE_CONTEXT_ISOLATION_VIOLATION`,
+`ISOLATION_EVIDENCE_STALE`, `ISOLATION_EVIDENCE_INVALID`.
+
+#### Strict runtime proof (US-0056 / DEC-0038)
+
+Per-phase isolation also requires strict runtime attestation tuples at
+boundaries (not artifact fields alone):
+
+- `orchestrator_run_id`, `runtime_proof_id`, `phase_id`, `role`
+- `proof_issued_at`, `proof_ttl_seconds`, `proof_hash`
+
+Fail-closed reason codes:
+`RUNTIME_PROOF_MISSING`, `RUNTIME_PROOF_INVALID`, `RUNTIME_PROOF_REUSED`,
+`RUNTIME_PROOF_STALE`, `RUNTIME_PROOF_AMBIGUOUS_LINK`.
+
+`/auto`, `/verify-work`, and `/release` must validate these tuples before
+continuation/finalization.
+
+#### `/auto` phase→role enforcement (US-0069 / DEC-0051)
+
+`/auto` uses a deterministic **phase→role matrix** plus scratchpad alternates
+(`AUTO_ROLE_RESEARCH`, `AUTO_ROLE_PLAN_VERIFY`, `AUTO_ROLE_REFRESH_CONTEXT`).
+Before each phase spawn it runs a **preflight capability gate**; missing
+capability stops with `PHASE_ROLE_CAPABILITY_MISSING` (no unrelated-role
+substitution). After each phase, isolation `role` and strict-proof `role` must
+match the same expected role or the run stops with `PHASE_ROLE_MISMATCH`.
+`execute` defaults to `dev`; non-`dev` requires
+`AUTO_EXECUTE_ROLE_OVERRIDE=allowed_non_dev_execute` **and**
+`EXECUTE_OVERRIDE_GOVERNANCE_REF` pointing to a parseable approved waiver. See
+`docs/engineering/runbook.md` and `decisions/DEC-0051.md`.
+
+#### `/auto` phase selection policy (US-0070 / DEC-0052)
+
+`/auto` builds a **resolved phase plan** from scratchpad before spawning phases:
+exactly one of `AUTO_PHASE_PLAN` (default `full`), `AUTO_PHASE_EXCLUDE`,
+`AUTO_PHASE_INCLUDE`, or `AUTO_PHASE_PROFILE` applies; conflicting selectors
+stop with `PHASE_POLICY_CONFLICT`. Non-skippable safety gates (`qa`,
+`verify-work`, `release`) and evidence-chain closure reinstate omitted phases
+with breadcrumb reasons such as `non_skippable_gate`. `start-from` and resume
+anchors **intersect** with the plan (`START_FROM_PHASE_PLAN_EMPTY_INTERSECTION`
+when empty). Backlog-drain, bulk execute, and team-mode runs **recompute** the
+plan each boundary. See `/auto`, `docs/engineering/runbook.md`, and
+`decisions/DEC-0052.md`.
+
+### Lightweight interaction
+
+Use `/ask` when you want to query the project without triggering the workflow:
+
+- "What's the current sprint status?"
+- "Which stories are still open?"
+- "How does the upgrade mode work?"
+- "What decision was made about X?"
+
+`/ask` reads the project artifacts (state, backlog, architecture, decisions, sprint
+progress) and answers from them. It never creates or modifies files. If your question
+reveals a bug or feature idea, it will suggest running `/intake`.
+
+### Memory drift auditing
+
+Use `/memory-audit` to check whether project memory artifacts still match
+repository reality. This is a read-only, non-blocking command that produces an
+advisory report at `docs/engineering/memory-drift-report.md`.
+
+**When to run:**
+
+- **Pre-handoff**: before writing any role handoff artifact.
+- **Pre-QA**: before `/qa` or `/verify-work`.
+- **Pre-release**: before `/release`.
+- **Ad-hoc**: after external code changes, long pauses, or whenever artifacts
+  feel stale.
+
+**How to interpret output:**
+
+The report contains a severity summary (`high` / `medium` / `low`) and a
+findings table with concrete evidence for each inconsistency. High-severity
+findings should be resolved before the next handoff or release. Medium and low
+findings can be addressed during `/refresh-context` or the next sprint.
+
+The report also includes a reference-only "Template drift" section. Template
+drift remediation belongs to US-0017 — `/memory-audit` only flags it for
+awareness.
+
+**Follow-up commands:**
+
+- `/refresh-context` — update stale artifacts.
+- `/sprint-plan` — if new work is discovered.
+- `/verify-work` — if acceptance status needs re-validation.
+- `/intake` — if findings reveal a new story or bug.
+
+### Workflow diagrams
+
+```mermaid
+flowchart TD
+  Intake[/intake/] --> Discovery[/discovery/]
+  Discovery --> Research[/research/]
+  Research --> Architecture[/architecture/]
+  Architecture --> SprintPlan[/sprint-plan/]
+  SprintPlan --> PlanVerify[/plan-verify/]
+  PlanVerify --> Execute[/execute/]
+  Execute --> QA[/qa/]
+  QA -->|fixes needed| Execute
+  QA --> VerifyWork[/verify-work/]
+  VerifyWork --> Release[/release/]
+  Release --> Refresh[/refresh-context/]
+  Execute --> Pause[/pause/]
+  QA --> Pause
+  Release --> Pause
+  Pause --> Resume[/resume/]
+  Resume --> Execute
+```
+
+```mermaid
+flowchart TD
+  Start[Idea] --> Intake2[/intake/]
+  Intake2 -->|DecisionGate| Decision{DecisionRequired}
+  Decision -->|ChooseOption| Discovery2[/discovery/]
+  Decision -->|ChooseOption| Research2[/research/]
+  Decision -->|ChooseOption| Architecture2[/architecture/]
+  Decision -->|ChooseOption| SprintPlan2[/sprint-plan/]
+  Decision -->|ChooseOption| PlanVerify2[/plan-verify/]
+  Decision -->|ChooseOption| Execute2[/execute/]
+  Decision -->|ChooseOption| QA2[/qa/]
+  Decision -->|ChooseOption| VerifyWork2[/verify-work/]
+  Decision -->|ChooseOption| Release2[/release/]
+  Decision -->|ChooseOption| Refresh2[/refresh-context/]
+  Decision -->|ChooseOption| Pause2[/pause/]
+  Discovery2 --> Research2
+  Research2 --> Architecture2
+  Architecture2 --> SprintPlan2
+  SprintPlan2 --> PlanVerify2
+  PlanVerify2 --> Execute2
+  Execute2 --> QA2
+  QA2 -->|fixes needed| Execute2
+  QA2 --> VerifyWork2
+  VerifyWork2 --> Release2
+  Release2 --> Refresh2
+  Execute2 --> Pause2
+  QA2 --> Pause2
+  Release2 --> Pause2
+  Pause2 --> Resume2[/resume/]
+  Resume2 --> Execute2
+  Execute2 --> HandoffDevQA[handoffs/dev_to_qa.md]
+  QA2 --> HandoffQAD[handoffs/qa_to_dev.md]
+  Intake2 --> HandoffPOTL[handoffs/po_to_tl.md]
+  SprintPlan2 --> HandoffTLDev[handoffs/tl_to_dev.md]
+```
+
+### Automation modes
+
+Configure in `.cursor/scratchpad.md`:
+
+- `AUTO_FLOW_MODE=manual|auto_until_decision`  
+  - `manual`: you trigger each phase/command yourself.  
+  - `auto_until_decision`: `/auto` continues by spawning fresh phase subagents until a decision gate, blocker, or pause boundary.
+- `PHASE_MODE=interactive|auto`  
+  - `interactive`: agent asks clarifying questions more often.  
+  - `auto`: agent minimizes prompts and proceeds with best effort.
+- `PERMISSION_MODE=interactive|auto`  
+  - `interactive`: ask before routine actions.  
+  - `auto`: reduce routine permission prompts.
+- `RUN_TESTS_ON_EDIT=0|1`  
+  - `1`: runs configured tests after meaningful edits.  
+  - `0`: tests only when you explicitly run QA/test phases.
+- `LOOP_UNTIL_GREEN=0|1`  
+  - `1`: keep iterating fix -> test until green (bounded).  
+  - `0`: run one pass and report failures.
+- `AUTO_IMPLEMENTATION_LOOP=0|1`  
+  - `1`: enables execute -> QA -> execute loop automatically with new Dev/QA subagent instances on each cycle.
+- `AUTO_LOOP_MAX_CYCLES=<n>`  
+  - safety cap for auto loops (recommended `3-7`, default `5`).
+- `AUTO_PAUSE_REQUEST=0|1`  
+  - `1`: request graceful stop at next safe boundary.
+- `AUTO_PAUSE_POLICY=after_task|after_phase`  
+  - `after_task`: faster stop, more frequent boundaries.  
+  - `after_phase`: cleaner checkpoints, fewer interruptions.
+
+### Sync policy (US-0038)
+
+Phase-triggered sync is policy-controlled and safe by default.
+
+Scratchpad controls:
+
+- `SYNC_POLICY_MODE=disabled|manual|by_phase|by_milestone|custom_phase_list`
+- `SYNC_CUSTOM_PHASES=<comma-separated canonical phases>`
+- `ALLOW_AUTO_PUSH=0|1`
+- `AUTO_PUSH_BRANCH_ALLOWLIST=<comma-separated branches/patterns>`
+
+Default-safe behavior:
+
+- Default mode is `manual` with `ALLOW_AUTO_PUSH=0` (no automatic push).
+- `disabled` and `manual` add near-zero overhead and preserve manual workflows.
+- Sync policy is evaluated only at completed phase boundaries.
+
+Guarded auto-push conditions (all must pass):
+
+1. Boundary matches configured mode.
+2. Auto-push is explicitly enabled (`ALLOW_AUTO_PUSH=1`).
+3. QA-first safety holds (feature work cannot auto-push pre-QA).
+4. No unresolved blocking QA findings/critical issues.
+5. Branch safety holds (protected/default branches denied unless allowlisted).
+6. Check chain passes (`TEST_COMMAND` required; optional lint/typecheck only if configured).
+
+Deterministic reason codes include:
+`SYNC_DISABLED`, `MANUAL_MODE_NO_AUTO`, `PRE_QA_AUTOPUSH_FORBIDDEN`,
+`BLOCKING_QA_FINDINGS`, `BRANCH_NOT_ALLOWLISTED`, `TEST_COMMAND_MISSING`,
+`TEST_FAILED`, `TEST_TIMEOUT`, `OPTIONAL_CHECK_FAILED`, `SYNC_PUSHED`.
+
+### Full scratchpad reference (detailed)
+
+- `MAGIC_CONTEXT_STRICT=0|1`  
+  - `1`: enforces context refresh discipline after code edits.
+- `DONE=0|1`  
+  - `1`: stop hook reminder loops when session is complete.
+- `MAGIC_BENCH_SESSION=<id>`  
+  - enables live benchmark event logging under one session id.
+- `AUTO_INSTALL_DEPS=0|1`  
+  - `1`: agent may install dependencies/runtimes automatically.
+- `AUTO_RELEASE_NOTES=0|1`  
+  - `1`: auto-generate `handoffs/release_notes.md`.
+- `REMOTE_EXECUTION=0|1`  
+  - `1`: allow remote/docker execution if configured.
+- `REMOTE_CONFIG=.cursor/remote.json`  
+  - path to remote execution server config.
+
+### Remote execution config (`.cursor/remote.json`)
+
+Remote config is optional and mode-aware:
+
+- `REMOTE_EXECUTION=0` (default): skip remote config checks entirely.
+- `REMOTE_EXECUTION=1`: validate `.cursor/remote.json` first and fail fast on
+  missing/malformed/invalid or insecure config.
+
+Canonical contract (DEC-0016):
+
+- Required root fields:
+  - `version` (integer)
+  - `defaultTarget` (string)
+  - `targets` (array)
+- Required target fields:
+  - `id` (string)
+  - `type` (`docker|ssh|vm`)
+  - `enabled` (boolean)
+  - `host` (string)
+  - `port` (integer `1..65535`)
+  - `workspaceRoot` (string)
+- Optional:
+  - `auth.mode` (`none|env`)
+  - If `auth.mode=env`, use env-var references only (`tokenEnv`,
+    `passwordEnv`, `privateKeyPathEnv`, ...).
+
+Two safe target examples are shipped in:
+
+- `.cursor/remote.json` (active repo)
+- `template/.cursor/remote.json` (template parity copy)
+
+The examples include:
+
+- `local-docker`: local network/docker-like endpoint.
+- `remote-vm-ssh`: remote VM/SSH-like endpoint.
+
+No secrets policy:
+
+- Never commit inline tokens/passwords/private keys in `remote.json`.
+- Commit env-var reference names only.
+
+Fail-fast error format:
+
+- `[REMOTE_CONFIG_ERROR] <path>: expected <rule>, got <actual>. Fix: <hint>.`
+
+Troubleshooting quick guide:
+
+- Missing file in remote mode:
+  - Create `.cursor/remote.json` from the template copy, or set
+    `REMOTE_EXECUTION=0`.
+- Invalid enum/type/range:
+  - Update the failing field to match allowed values/ranges.
+- Malformed JSON:
+  - Fix JSON syntax and retry.
+- Secret-like inline value detected:
+  - Replace literal secret with an env-var reference field.
+- **CI still runs its-magic packaging jobs?** Your project received a pre-fix workflow.
+  Run **`its-magic --target <repo> --mode upgrade`** (or **`--mode clean`** then reinstall)
+  to refresh `.github/workflows/ci.yml` from the corrected template. After upgrade, GitHub
+  Actions should show only **`checks`** and **`auto-fix`** jobs — not `npm-test`,
+  `brew-test`, or `choco-test`. Fix applies to new installs/upgrades; stale repos heal on
+  next upgrade (**US-0018**).
+
+Team/local (recommended in `.cursor/scratchpad.local.md`):
+
+- `TEAM_MODE=0|1`
+- `TEAM_MEMBER=<your-id>`
+- `ACTIVE_TASK_IDS=T-12,T-13`
+
+### Automated feature loop (optional)
+
+Enable:
+
+- `AUTO_FLOW_MODE=auto_until_decision`
+- `PHASE_MODE=auto`
+- `PERMISSION_MODE=auto`
+- `RUN_TESTS_ON_EDIT=1`
+- `LOOP_UNTIL_GREEN=1`
+- `AUTO_IMPLEMENTATION_LOOP=1`
+- `AUTO_LOOP_MAX_CYCLES=5`
+
+Then run `/auto`.
+
+Graceful stop (for shutdown/end of day):
+
+1. Set `AUTO_PAUSE_REQUEST=1`
+2. Flow stops at next configured boundary (`AUTO_PAUSE_POLICY`)
+3. `/pause` artifacts are written
+4. Next day run `/resume` or `/auto`
+
+### Recommended profiles
+
+**Max automation (high autonomy):**
+
+- `AUTO_FLOW_MODE=auto_until_decision`
+- `PHASE_MODE=auto`
+- `PERMISSION_MODE=auto`
+- `RUN_TESTS_ON_EDIT=1`
+- `LOOP_UNTIL_GREEN=1`
+- `AUTO_IMPLEMENTATION_LOOP=1`
+- `AUTO_LOOP_MAX_CYCLES=5`
+- `AUTO_INSTALL_DEPS=1` (optional, if you trust auto installs)
+- `AUTO_PAUSE_POLICY=after_phase`
+
+**Safer automation (recommended for most teams):**
+
+- same as above, but keep:
+  - `PERMISSION_MODE=interactive`
+  - `AUTO_INSTALL_DEPS=0`
+  - `AUTO_PAUSE_POLICY=after_task`
+
+### Quality chain (3-layer auto-fix)
+
+its-magic provides a complete quality chain that catches issues at three levels.
+Each layer catches problems the previous layer missed:
+
+```text
+┌─────────────────────────────────────────────────────────────────┐
+│ Layer 1: Cursor AI loop (in-editor)              OFF by default │
+│   AUTO_IMPLEMENTATION_LOOP + LOOP_UNTIL_GREEN                   │
+│   execute → QA → fix → execute (bounded by AUTO_LOOP_MAX_CYCLES)│
+└──────────────────────────┬──────────────────────────────────────┘
+                           │ code ready to push
+┌──────────────────────────▼──────────────────────────────────────┐
+│ Layer 2: validate-and-push (local pre-push)      MANUAL (run it)│
+│   scripts/validate-and-push.sh / .ps1                           │
+│   test → format → lint-fix → test → commit + push               │
+└──────────────────────────┬──────────────────────────────────────┘
+                           │ pushed to GitHub
+┌──────────────────────────▼──────────────────────────────────────┐
+│ Layer 3: CI auto-fix (GitHub Actions)            OFF by default │
+│   .github/workflows/ci.yml                                      │
+│   test/lint → auto-fix → commit → re-run (up to 3 retries)     │
+└─────────────────────────────────────────────────────────────────┘
+```
+
+| Layer | Default | Enable |
+|-------|---------|--------|
+| 1 - Cursor AI loop | off | Set `AUTO_IMPLEMENTATION_LOOP=1` + `LOOP_UNTIL_GREEN=1` in scratchpad |
+| 2 - validate-and-push | manual | Run `scripts/validate-and-push.sh` or `.ps1` before pushing |
+| 3 - CI auto-fix | off | Set `CI_AUTO_FIX: true` in `docs/engineering/runbook.md` |
+
+CI itself (tests, lint, typecheck) always runs on push/PR. Only the **auto-fix
+retry loop** is gated behind `CI_AUTO_FIX`. When disabled, CI still reports
+failures -- it just won't try to fix and commit automatically.
+
+All commands are read from `docs/engineering/runbook.md`. Fill in your
+project-specific commands once and every layer uses them:
+
+```text
+TEST_COMMAND: npm test
+LINT_COMMAND: npx eslint .
+LINT_FIX_COMMAND: npx eslint --fix .
+FORMAT_COMMAND: npx prettier --write .
+CI_AUTO_FIX: true
+```
+
+#### Layer 1: Cursor AI loop
+
+Enabled via scratchpad flags (see [Automation modes](#automation-modes)).
+The AI runs execute → QA → fix cycles inside Cursor until tests pass or
+the safety cap (`AUTO_LOOP_MAX_CYCLES`) is reached.
+
+#### Layer 2: Local validate-and-push
+
+Run before pushing to catch anything the AI loop missed. **Merged scratchpad** (see
+`docs/engineering/runbook.md`, **Executable validate-and-push wiring (DEC-0058)**) gates
+**`git push`**: default **`SYNC_POLICY_MODE=manual`** and **`ALLOW_AUTO_PUSH=0`** exit early
+with a **reason code** (no push). Opt-in push requires an eligible mode, **`ALLOW_AUTO_PUSH=1`**,
+a non-empty **branch allowlist** match, passing **runbook** checks, and bounded **QA** rules.
+
+```bash
+# Bash (Linux / macOS; bash required for this script)
+bash scripts/validate-and-push.sh
+
+# PowerShell (Windows)
+powershell scripts/validate-and-push.ps1
+powershell scripts/validate-and-push.ps1 -MaxAttempts 3
+powershell scripts/validate-and-push.ps1 -DryRun
+```
+
+The script:
+1. Evaluates merged scratchpad policy via **`python scripts/sync_push_gates.py`** (Python 3 on PATH)
+2. Runs `FORMAT_COMMAND` and `LINT_FIX_COMMAND` to auto-fix what it can
+3. Runs `LINT_COMMAND`, optional `TYPECHECK_COMMAND`, and `TEST_COMMAND` to verify (with `TEST_TIMEOUT_SECONDS` when `timeout`/`gtimeout` is available on Unix)
+4. If checks fail, pauses and waits for you to fix
+5. Re-runs (up to 5 attempts, configurable)
+6. When green, re-checks allowlist + QA scan, then commits and pushes automatically (unless dry-run / no-commit)
+
+Use `-NoCommit` (PowerShell), **`--dry-run`** first arg (Bash), or `false` as third arg (Bash) to skip **push**.
+**Policy-only** interpretation of scratchpad sync flags is **deprecated** for these scripts; see **`decisions/DEC-0058.md`** (policy semantics remain **`DEC-0018`** / **`US-0038`**).
+
+#### Layer 3: CI auto-fix (GitHub Actions)
+
+**Disabled by default.** Set `CI_AUTO_FIX: true` in `docs/engineering/runbook.md`
+to enable. When enabled and CI fails after a push, the auto-fix job kicks in:
+
+```text
+push / PR  ──>  checks  ──>  PASS  ──>  done
+                   │
+                  FAIL
+                   │
+             auto-fix job
+                   │
+          run LINT_FIX_COMMAND
+          run FORMAT_COMMAND
+                   │
+             changes found?
+            ╱              ╲
+         yes                no
+          │                  │
+    commit + push       report failure
+          │             (manual fix needed)
+     CI re-runs
+     (up to 3x)
+```
+
+Auto-fix commits appear as `ci: auto-fix attempt N/3`. After 3 retries the
+workflow stops and points you to `scripts/validate-and-push` for local fixing.
+
+<!-- readme-feature-coverage-catalog -->
+
+### Feature coverage catalog (US-0091)
+
+- `/acceptance` — Mandatory Intake Question Packs for First and Small Intakes (`US-0068`).
+- `/ask` — /ask Command: Context-Aware Questions Without Workflow (`US-0020`).
+- `/ask` — Context Compaction and Tiered Token-Cost Optimization Mode (`US-0053`).
+- `/auto` — Architecture triad archiver ignores `## US-xxxx` headings, blocking `/auto` with `STATE_ARCHIVE_BOUNDARY_AMBIGUOUS` (`BUG-0010`).
+- `/auto` — Configurable Auto Phase Selection Policy (`US-0070`).
+- `/auto` — Continuous `/auto` Backlog-Drain Mode with Fine-Tune Switches (`US-0044`).
+- `/auto` — Fresh Subagent Context Per Phase and /auto Orchestration (`US-0023`).
+- `/auto` — Mid-Process `/auto` Continuation with Deterministic Resume Point (`US-0037`).
+- `/auto` — Strict Phase Role Enforcement in /auto Orchestration (`US-0069`).
+- `/auto` — Strict Runtime Proof for Per-Phase Subagent Isolation (`US-0056`).
+- `/auto` — Token-Cost Hardening for Orchestrated Runs (`US-0080`).
+- `/auto` — `/auto` continuous multi-phase loop + quiet drain (close one-phase-stop gap) (`US-0088`).
+- `/auto` — `/auto` executes phases without spawning required subagents (`BUG-0006`).
+- `/auto` — `/auto` explicit bug targeting (fix all OPEN bugs / fix `BUG-####`) (`US-0087`).
+- `/auto` — `/auto` fails with stale resume target after bug intake (`BUG-0005`).
+- `/check` — Optional Documentation Pack (Design Concept, CRS, Technical Spec) (`US-0031`).
+- `/confirmation` — Enforced Interactive Intake Question Evidence (`US-0078`).
+- `/connectivity` — Release Operator Run/Connect/Verify Hints Contract (`US-0067`).
+- `/decision` — Optional Fresh-Project ID Namespace Bootstrap (`US-0052`).
+- `/derived` — Deterministic Status Reconciliation Command (`US-0055`).
+- `/developer-dense` — Documentation Audience Profiles and Dual README Strategy (`US-0077`).
+- `/docs` — Installer-Owned `its_magic/` Folder for Framework Metadata (`US-0062`).
+- `/engineering` — Deterministic Context Slimming and Archive Enforcement Across Core Artifacts (`US-0072`).
+- `/engineering` — Deterministic State Hot-Surface Rollover and Archive Enforcement (`US-0060`).
+- `/engineering` — OS-Aware Runbook Command Auto-Bootstrap with Verified Quality Gates (`US-0063`).
+- `/flag` — Release Gate for Command/Flag Documentation Delta (`US-0030`).
+- `/intake` — Critical Evaluation in Intake and Architecture (`US-0021`).
+- `/intake` — Deterministic Intake Runtime Capability Guard and Single-Writer Drift Safety (`US-0059`).
+- `/intake` — Multi-Repo and Contract Compatibility Observability (`US-0034`).
+- `/intake` — intake evidence records asked questions that were never asked (`BUG-0007`).
+- `/integration` — Generated Test Scaffolding and Auto-Run Contract (`US-0066`).
+- `/managed` — Runtime QA Autopilot for Generated Projects (`US-0065`).
+- `/new` — First-Intake Full-Plan Coverage and Story-Map Gate (`US-0081`).
+- `/order` — Deterministic Artifact Ordering and Write Discipline (`US-0058`).
+- `/phases` — Cross-Phase Artifact Ownership Guard and Deterministic Archive Control (`US-0061`).
+- `/planning` — User-Visible Internal Metadata Sanitization Guard (`US-0071`).
+- `/product` — Backlog Reconciliation Gate for Released Sprints (`US-0043`).
+- `/push` — Phase-Triggered Sync Policy with Guarded Auto-Push (`US-0038`).
+- `/release` — Enforced Per-Phase Subagent Isolation with Audit Gate (`US-0048`).
+- `/release` — Legacy DONE-Story Acceptance/Traceability Backfill Guard (`US-0049`).
+- `/release` — Per-Sprint Release Notes and Release Queue Tracker (`US-0040`).
+- `/release` — Release Findings Artifact and Post-QA Issue Workflow (`US-0042`).
+- `/release` — Release Gate Tightening for Check-In Tests and QA/UAT Completion (`US-0039`).
+- `/remote` — Automation-driven remote execution selection (Docker / SSH / NL container intent) (`US-0086`).
+- `/remote` — Gitignored `.env` for remote and release connectivity (no AI read) (`US-0085`).
+- `/repetitive` — Delegable Intake Clarification Without Hard Blocks (`US-0083`).
+- `/research` — Knowledge Curation & Early Research (`US-0029`).
+- `/risk` — Intelligent Intake Decomposition and Risk-Aware PO Questioning (`US-0051`).
+- `/scratchpad` — Caveman mode missing voice compression rules (US-0089 incomplete delivery) (`BUG-0011`).
+- `/scratchpad` — Executable Scratchpad-Driven Sync and Auto-Push Wiring (`US-0076`).
+- `/scratchpad` — Scratchpad Delivery Simplification (Example-Only Install Policy) (`US-0073`).
+- `/scratchpad` — Upgrade Scratchpad Example–First Refresh (Fix Example Drift vs Materialized Baseline) (`US-0075`).
+- `/scratchpad` — Upgrade-Safe Scratchpad Example Refresh and Parity (`US-0057`).
+- `/sprint-plan` — Explicit `/sprint-plan --bulk` Mode (`US-0046`).
+- `/sprint-plan` — Sprint Sizing Rules and Configurable Sprint Planning (`US-0022`).
+- `/story` — Optional Feature User Guide Generation (`US-0032`).
+- `/uat` — Cursor browser-integrated UAT self-test (browser_smoke + automatable manual UI) (`US-0093`).
+- `/uat` — UAT Artifact Lifecycle and Ownership (`US-0027`).
+- `SKILL` scratchpad flag — Skill and Templates (`US-0004`).
+- `US-0001` scratchpad flag — Core Workflow Commands (`US-0001`).
+
+## Walkthrough examples
+
+### Example 1: New feature from idea
+
+1. `/intake`
+2. `/research`
+3. `/architecture`
+4. `/sprint-plan`
+5. `/plan-verify`
+6. `/execute`
+7. `/qa`
+8. `/verify-work`
+9. `/release`
+10. `/refresh-context`
+
+### Example 2: Mid-flight idea change
+
+1. Set `AUTO_PAUSE_REQUEST=1`
+2. Run `/intake` to update story/acceptance
+3. Re-run `/sprint-plan` + `/plan-verify`
+4. Resume via `/auto`
+
+### Example 3: Pause/resume
+
+1. `/pause`
+2. Close work
+3. `/resume` next session
+
+### Deterministic `/auto` continuation
+
+When resuming mid-process, `/auto` resolves start phase deterministically:
+
+1. explicit `/auto start-from=<phase>`
+2. `handoffs/resume_brief.md`
+3. conservative `docs/engineering/state.md` fallback
+4. fail-fast (no guessing)
+
+Canonical phases:
+`intake`, `discovery`, `research`, `architecture`, `sprint-plan`,
+`plan-verify`, `execute`, `qa`, `verify-work`, `release`, `refresh-context`.
+
+Fail-fast message format:
+`[AUTO_RESUME_ERROR] <code>: <summary>. Source=<source>. Fix: <action>.`
+
+Compatibility and safety:
+- Manual/interactive workflow stays unchanged unless `/auto` continuation is used.
+- Existing stop conditions remain enforced (decision gate, missing input,
+  pause request, loop max).
+
+### Optional `/auto` backlog-drain mode (US-0044)
+
+If you want `/auto` to continue across multiple planned stories in one run,
+enable backlog-drain switches in `.cursor/scratchpad.md`:
+
+- `AUTO_BACKLOG_DRAIN=1`
+- `AUTO_BACKLOG_MAX_STORIES=<n>`
+- `AUTO_BACKLOG_ON_BLOCK=stop|skip`
+- `AUTO_STORY_SELECTION=priority_then_backlog_order`
+
+Default-safe behavior remains unchanged with `AUTO_BACKLOG_DRAIN=0`.
+
+### Explicit `/sprint-plan --bulk` mode (US-0046)
+
+By default, `/sprint-plan` plans one scope at a time. For multi-story planning,
+run explicit bulk mode:
+
+- `/sprint-plan --bulk`
+
+Bulk planning remains bounded and deterministic via `.cursor/scratchpad.md`:
+
+- `SPRINT_BULK_MAX_STORIES=<n>`
+- `SPRINT_BULK_MAX_SPRINTS=<n>`
+- `SPRINT_BULK_SELECTION=priority_then_backlog_order`
+
+Bounded stop reason codes:
+`SPRINT_BULK_MAX_STORIES_REACHED`, `SPRINT_BULK_MAX_SPRINTS_REACHED`,
+`SPRINT_BULK_NO_ELIGIBLE_STORIES`, `SPRINT_BULK_MISSING_ACCEPTANCE`.
+
+### Explicit `/auto --execute-bulk` mode (US-0047)
+
+Bulk execution is explicit-mode only. Default `/auto` behavior remains unchanged.
+
+Enable either way:
+
+- one-run explicit argument: `/auto --execute-bulk`
+- scratchpad switch: `AUTO_EXECUTE_BULK=1`
+
+Deterministic controls in `.cursor/scratchpad.md`:
+
+- `AUTO_EXECUTE_MAX_ITEMS=<n>`
+- `AUTO_EXECUTE_ON_BLOCK=stop|skip`
+- `AUTO_EXECUTE_SELECTION=planned_then_priority`
+- `AUTO_TEAM_SCOPE_ENFORCE=0|1`
+
+Deterministic reason codes:
+`EXEC_BULK_MAX_ITEMS_REACHED`, `EXEC_BULK_NO_ELIGIBLE_ITEMS`,
+`EXEC_BULK_ITEM_BLOCKED_STOP`, `EXEC_BULK_ITEM_BLOCKED_SKIPPED`,
+`EXEC_TEAM_SCOPE_BLOCKED`, `EXEC_TEAM_SCOPE_SKIPPED`.
+
+Team-mode safety:
+- In `TEAM_MODE=1`, bulk execute records `TEAM_MODE`, `TEAM_MEMBER`,
+  `ACTIVE_TASK_IDS` in state breadcrumbs.
+- With `AUTO_TEAM_SCOPE_ENFORCE=1`, out-of-scope tasks are blocked/skipped
+  deterministically and never mutated.
+
+### Example 4: Existing project onboarding
+
+1. `/map-codebase`
+2. Review generated mapping artifacts
+3. Continue with `/intake` or `/architecture`
+
+## Other useful capabilities
+
+### Voice input (multilingual)
+
+Voice is an input layer only; it feeds normal slash commands.
+
+- OS dictation
+- Cursor voice (if available)
+- Local STT tooling
+
+Reliable pattern:
+
+- bind `/intake ` insertion shortcut
+- dictate only the content after the command
+
+### Repository layout (quick orientation)
+
+- `.cursor/`: commands, rules, agents, hooks, skills, scratchpad.
+- `docs/`: product + engineering docs.
+- `sprints/`: sprint planning/tracking.
+- `handoffs/`: role-to-role transfers.
+- `decisions/`: decision records.
+- `.github/workflows/`: CI/CD templates.
+
+<!-- readme-feature-coverage-catalog -->
+
+### Feature coverage catalog (US-0091)
+
+- `/evidence` — Backlog-to-Sprint Traceability Contract (`US-0025`).
+- `/exit` — Milestone Lifecycle Definition and Exit Criteria (`US-0026`).
+- `/field` — Official Remote Config Template, Docs, and Fail-Fast Validation (`US-0036`).
+- `/installer` — Runbook Completion (`US-0015`).
+- `/intake` — Component-Scoped Execution Mode with Protection Guards (`US-0035`).
+- `/intake` — Configurable Guided Intake Behavior (`US-0033`).
+- `/map-codebase` — map-codebase does not write codebase-map in fresh repos (`BUG-0002`).
+- `/memory-audit` — Memory Drift Audit Command (`US-0024`).
+- `/product` — Canonical Story Status Source + Global Drift Guard (`US-0045`).
+- `/product` — Clean Install Hygiene and Complete Clean-Repo Coverage (`US-0050`).
+- `/security-review` — Security & Compliance Review Agent (`US-0028`).
+- `/skip` — Explicit Bulk Execute Orchestration Mode (`US-0047`).
+- `/strings` — Clean Placeholder Content from Templates and Active Files (`US-0019`).
+- `/write` — Artifact Templates and Starter Docs (`US-0006`).
+- `AUTO_FLOW_MODE` scratchpad flag — Automation Modes (`US-0011`).
+- `README` scratchpad flag — Voice Input Documentation (`US-0010`).
+- `TEAM_MODE` scratchpad flag — Team Mode (`US-0013`).
+- `US-0002` scratchpad flag — AI Behavior Rules (`US-0002`).
+- `US-0003` scratchpad flag — Subagent Definitions (`US-0003`).
+- `US-0005` scratchpad flag — Hook System (`US-0005`).
+- `US-0008` scratchpad flag — CLI Installer (`US-0008`).
+- `US-0012` scratchpad flag — Benchmark Suite (`US-0012`).
+- `US-0014` scratchpad flag — Quality Chain (3-Layer) (`US-0014`).
+- `US-0017` scratchpad flag — Template Drift Guard (`US-0017`).
+
+## Developer and release deep-dive
+
+### CI/CD via runbook
+
+Workflows read keys from `docs/engineering/runbook.md`:
+
+- `TEST_COMMAND`
+- `LINT_COMMAND`
+- `TYPECHECK_COMMAND`
+- `DEPLOY_STAGING_COMMAND`
+- `DEPLOY_PROD_COMMAND`
+
+Unset keys are skipped. The template ships with empty values for `LINT_COMMAND`,
+`FORMAT_COMMAND`, and `TYPECHECK_COMMAND` -- this is intentional. its-magic is a
+template/installer project; fill in your project-specific commands after setup.
+
+US-0015 intent contract:
+- Empty optional runbook keys are valid defaults for this repository type.
+- They must not be treated as missing required configuration.
+
+### Installer internals
+
+- `installer.ps1` (Windows)
+- `installer.sh` (macOS/Linux)
+- `installer.py` (fallback)
+
+Modes: `missing`, `overwrite`, `interactive`, `upgrade` (+ optional backup).
+
+### Release automation
+
+Unified release scripts:
+
+- Windows: `scripts/release-all.ps1`
+- macOS/Linux: `scripts/release-all.sh`
+
+NPM helpers:
+
+- `npm run release:all`
+- `npm run release:all:patch|minor|major|beta|dry`
+- `npm run release:npm-only|choco-only|brew-only`
+
+Release script flow:
+
+1. bump `package.json` version
+2. publish npm
+3. create GitHub release
+4. update/publish Chocolatey package
+5. update/push Homebrew formula (stable or beta)
+
+```mermaid
+flowchart LR
+  ReleaseAll[scripts/release-all.*] --> VerCheck{version has -?}
+  VerCheck -->|stable| NPM["npm publish --tag latest"]
+  VerCheck -->|prerelease| NPMBeta["npm publish --tag beta"]
+  ReleaseAll --> GH["gh release create"]
+  VerCheck -->|prerelease| GHPre["--prerelease flag"]
+  GH --> Choco[choco pack + push]
+  GH --> BrewCheck{prerelease?}
+  BrewCheck -->|no| BrewStable[its-magic.rb]
+  BrewCheck -->|yes| BrewBeta[its-magic-beta.rb]
+  NPM --> U1["npx its-magic"]
+  NPMBeta --> U1b["npx its-magic@beta"]
+  Choco --> U2["choco install its-magic"]
+  Choco --> U2b["choco install its-magic --pre"]
+  BrewStable --> U3["brew install its-magic"]
+  BrewBeta --> U3b["brew install its-magic-beta"]
+```
+
+Prereqs:
+
+- `npm login`
+- `gh auth login`
+- Chocolatey API key (if choco publish)
+- Homebrew tap repo for formula distribution
+
+### Package manager installation matrix
+
+| Manager    | Stable                                    | Beta / Pre-release                          |
+|------------|-------------------------------------------|---------------------------------------------|
+| npm/npx    | `npx its-magic --target . --mode missing` | `npx its-magic@beta --target . --mode missing` |
+| Chocolatey | `choco install its-magic`                 | `choco install its-magic --pre`             |
+| Homebrew   | `brew install USER/tap/its-magic`         | `brew install USER/tap/its-magic-beta`      |
+
+### Release package contents
+
+Published npm package includes runtime content only (commands/rules/agents/docs/installers).
+
+Excluded from npm package:
+
+- `benchmarks/`
+- `tests/`
+- `packaging/`
+- `Plan.md`
+
+### Benchmarks
+
+- Main benchmark: `benchmarks/run-bench.ps1` or `benchmarks/run-bench.sh`
+- Live benchmark: `benchmarks/live/run-live-bench.*`
+- Prompted benchmark: `benchmarks/prompts/run-prompts.*`
+- Headless benchmark: `benchmarks/headless/run-headless.*`
+
+Reports:
+
+- `benchmarks/bench-report.md`
+- `benchmarks/live/live-bench-report.md`
+- `benchmarks/headless/headless-report.md`
+- `benchmarks/headless/protocol.md`
+
+```mermaid
+flowchart TD
+  StartBench[Start benchmark] --> SelectScenario[Load scenarios]
+  SelectScenario --> InstallKit[Install its-magic into temp workspace]
+  InstallKit --> RunChecks[Validate required files/sections]
+  RunChecks --> BenchReport[Write benchmarks/bench-report.md]
+```
+
+```mermaid
+flowchart TD
+  StartLive[Start live benchmark] --> SetSession[Set MAGIC_BENCH_SESSION]
+  SetSession --> RunCommands[Run /* commands in Cursor]
+  RunCommands --> LogHooks[Hook telemetry to bench-log.jsonl]
+  LogHooks --> LiveReport[Write live-bench-report.md]
+```
+
+```mermaid
+flowchart TD
+  StartHeadless[Start headless run] --> LoadPrompt[Load prompt blocks]
+  LoadPrompt --> TempWorkspace[Create temp workspace]
+  TempWorkspace --> InstallHeadless[Install its-magic]
+  InstallHeadless --> RunAgent[agent -p --force for each step]
+  RunAgent --> ValidateOutputs[Validate files/sections/smoke checks]
+  ValidateOutputs --> WriteProtocol[Write protocol.md]
+  WriteProtocol --> WriteHeadlessReport[Write headless-report.md]
+```
+
+### Rules
+
+- `core.mdc`: phase flow, context pack, pause/resume, remote usage.
+- `quality.mdc`: small steps, tests/quality, optional auto-install.
+- `coding-standards.mdc`: strict language best practices and code quality rules.
+- `handoffs.mdc`: handoffs + state updates required.
+- `escalation.mdc`: decision gate and stop conditions.
+
+### Hooks
+
+- `beforeShellExecution`: blocks dangerous commands.
+- `beforeReadFile`: warns on secret-like files.
+- `afterFileEdit`: tracks code edits vs context refresh.
+- `stop`: reminds context refresh when needed.
+
+### Artifacts (single source of truth)
+
+- `docs/product/*`: vision, backlog, acceptance.
+- `docs/engineering/*`: architecture, decisions, state, runbook.
+- `sprints/Sxxxx/*`: sprint scope, tasks, progress, QA findings, summary.
+- `decisions/*`: decision records.
+- `handoffs/*`: role-to-role transfer notes.
+
+## Purpose
+
+This repository publishes the **its-magic** workflow kit: commands, rules, skills, and
+documentation templates that teams install into their own repositories. The goal is a
+repeatable, file-backed lifecycle from intake through release.
+
+## Quickstart
+
+Use [Setup](#setup) for install commands. First-time install:
+
+```bash
+npx its-magic --target . --mode missing --create
+```
+
+## Examples
+
+- Upgrade an existing repo: `its-magic --target . --mode upgrade`
+- Run check-in tests: use `TEST_COMMAND` from `docs/engineering/runbook.md` (often `sh tests/run-tests.sh`).
+
+## Related documentation
+
+- Operator commands and gates: `docs/engineering/runbook.md`
+- Architecture and story contracts: `docs/engineering/architecture.md`
+- Product backlog and acceptance: `docs/product/backlog.md`, `docs/product/acceptance.md`
+- Optional spec-pack mode (`SPEC_PACK_MODE=1`): engineering design artifacts under `docs/engineering/` when your team enables it
+- Optional user guides (`USER_GUIDE_MODE=1`): `docs/user-guides/` when enabled
+
+## Limitations
+
+- its-magic is a **process and documentation** framework; it does not replace your
+  application runtime, hosting, or product-specific compliance work.
+- Mixed files such as `README.md` are preserved on upgrade; review notices may appear when
+  the template adds new sections.
+- Documentation profile validation (`scripts/validate_doc_profile.py`) enforces audience and
+  depth choices from the merged scratchpad (`DOC_AUDIENCE_PROFILE`, `DOC_DETAIL_LEVEL`).
+
+## Contributing
+
+Contributor-focused workflow and guardrails live in
+[`docs/developer/README.md`](docs/developer/README.md).