popilot 0.2.0 → 0.3.0

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Yoonjae Song
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -2,7 +2,7 @@
 
 > *"Developers have Copilot. Product Owners have Popilot."*
 
-**Popilot** is a multi-agent AI assistant for Product Owners and Product Managers, built on [Claude Code](https://docs.anthropic.com/en/docs/claude-code). It scaffolds a complete agent team — 14 specialized personas, 30+ slash commands, document templates, and workflow automation — so you can focus on product decisions while AI handles the heavy lifting.
+**Popilot** is a multi-agent AI assistant for Product Owners and Product Managers, built on [Claude Code](https://docs.anthropic.com/en/docs/claude-code). It scaffolds a complete agent team — 15 specialized personas, 30+ slash commands, document templates, and workflow automation — so you can focus on product decisions while AI handles the heavy lifting.
 
 ---
 
@@ -15,7 +15,7 @@ Popilot solves this with a **team of specialized agents**, each with deep expert
 - **No more context-switching** — Oscar routes your request to the right agent automatically
 - **Structured workflows** — From PRD to Screen Spec to Story to Dev Handoff, nothing falls through the cracks
 - **Data-driven decisions** — Danny queries your analytics; Vicky validates your hypotheses; Tara governs data quality
-- **Full PO coverage** — Strategy, market research, sprint planning, screen specs, tracking, operations, and more
+- **Full PO coverage** — Strategy, market research, GTM, sprint planning, screen specs, tracking, operations, and more
 - **Living specs, not dead docs** — spec-site turns static markdown into interactive, scenario-based mockups that developers actually use
 
 ---
@@ -33,13 +33,14 @@ Then open [Claude Code](https://docs.anthropic.com/en/docs/claude-code) and type
 
 ## The Agent Team
 
-### PO Team (8 + Orchestrator)
+### PO Team (10 + Orchestrator)
 
 | Icon | Name | Command | Role | Specialty |
 |------|------|---------|------|-----------|
 | :tophat: | **Oscar** | *(default)* | Orchestrator | Routing, coordination, proactive alerts |
 | :dart: | **Simon** | `/strategy` | Strategist | PRD, hypotheses, product strategy |
 | :world_map: | **Marco** | `/market` | Market Researcher | Competitive analysis, positioning |
+| :loudspeaker: | **Mia** | `/gtm` | GTM Strategist | Go-to-market strategy, launch planning, messaging |
 | :clipboard: | **Penny** | `/plan` | Planner | Sprint planning, stories, backlog |
 | :triangular_ruler: | **Hank** | `/handoff` | Handoff Specialist | Screen specs, dev handoff, release readiness |
 | :bar_chart: | **Vicky** | `/validate` | Validator | Hypothesis validation, guard rails, OKR tracking |
@@ -66,6 +67,7 @@ Then open [Claude Code](https://docs.anthropic.com/en/docs/claude-code) and type
 
 ```
 Strategy:   Simon ←→ Marco
+GTM:        Mia ←→ Simon + Marco
 Execute:    Penny → Hank → Derek + Quinn
 Insight:    Danny ←→ Rita
 Measure:    Vicky ←→ Tara
@@ -153,6 +155,7 @@ After setup, open Claude Code and type `/start`. Oscar can run a **deep intervie
 |---------|-------------|
 | `/strategy` | Activate Simon (strategy, PRD) |
 | `/market` | Activate Marco (competitive analysis, positioning) |
+| `/gtm` | Activate Mia (GTM strategy, launch planning, messaging) |
 | `/plan` | Activate Penny (sprint planning) |
 | `/validate` | Activate Vicky (hypothesis validation) |
 | `/analytics` | Activate Danny (data analysis) |
@@ -202,9 +205,9 @@ my-project/
 │   ├── .secrets.yaml                  # Sensitive data (gitignored)
 │   ├── WORKFLOW.md                    # Workflow guide (hydrated)
 │   │
-│   ├── agents/                        # 14 agent personas (+ TEMPLATE.md)
+│   ├── agents/                        # 15 agent personas (+ TEMPLATE.md)
 │   ├── oscar/workflows/               # Oscar extension modules
-│   ├── templates/                     # 11 document templates
+│   ├── templates/                     # 12 document templates
 │   ├── integrations/                  # Provider configs + registry
 │   ├── metrics/                       # Metrics data
 │   ├── daily/                         # Daily work logs
@@ -319,6 +322,7 @@ Set `spec_site.deploy_url` in `project.yaml` — agents will use this URL in han
 | Template | Owner | File | Purpose |
 |----------|-------|------|---------|
 | Sprint PRD | Simon | `templates/prd.md` | One Question, hypothesis, OMTM |
+| GTM Plan | Mia | `templates/gtm-plan.md` | ICP, messaging, channel mix, launch KPI gates |
 | Epic Spec | PO | `templates/epic-spec.md` | WHY, WHAT, HOW, edge cases |
 | Story v2 | Penny | `templates/story-v2.md` | AC (Given-When-Then), task breakdown |
 | Screen Spec | Hank | `templates/screen-spec.md` | 7-level UI specification |
@@ -339,7 +343,7 @@ popilot <command> [target-dir] [options]
 
 Commands:
   init [dir]      Scaffold + interactive setup + hydration (default)
-  hydrate [dir]   Re-hydrate .hbs templates from existing project.yaml
+  hydrate [dir]   Sync latest scaffold templates + re-hydrate from project.yaml
   doctor [dir]    Check installation health
   help            Show this help
 
@@ -349,13 +353,31 @@ Options:
 
 Examples:
   npx popilot init my-project
-  npx popilot hydrate
+  npx popilot@latest hydrate
+  npx popilot@latest hydrate --force
   npx popilot doctor
   npx popilot my-project          # same as: popilot init my-project
 ```
 
 ---
 
+## Upgrading Existing Projects
+
+For projects initialized with older Popilot versions:
+
+```bash
+# Safe upgrade (adds missing latest templates/files, then hydrates)
+npx popilot@latest hydrate
+
+# Full refresh (overwrites existing scaffold files, then hydrates)
+npx popilot@latest hydrate --force
+```
+
+- `hydrate` now syncs the latest scaffold before rendering.
+- Use `--force` only when you intentionally want to replace existing scaffold-managed files.
+
+---
+
 ## How It Works
 
 1. **`npx popilot init`** copies the scaffold and runs the setup wizard

package/adapters/claude-code/.claude/commands/_domain.md.hbs

@@ -10,6 +10,7 @@ Load {{name}} related context.
 
 3. Recommend a suitable agent:
    - 🎯 Simon: Strategy/PRD discussion
+   - 📣 Mia: GTM/launch/messaging planning
    - 📈 Danny: Data analysis
    - 📊 Vicky: Validation/analysis
    - 🎤 Rita: Customer insights

package/adapters/claude-code/.claude/commands/gtm.md

@@ -0,0 +1,82 @@
+# /gtm - Activate GTM Strategist
+
+Activate 📣 **Mia** (GTM Strategist) agent.
+
+## Load Persona
+
+Read `.context/agents/gtm-strategist.md` and activate Mia's persona.
+
+## Mia's Identity
+
+- **Role**: GTM Strategist + Launch Architect
+- **Personality**: Outcome-driven PMM who turns product value into adoption
+- **Strengths**: ICP focus, messaging design, launch sequencing, KPI gates
+
+## Communication Style
+
+- Strategic but execution-ready
+- Segment/message/channel structured outputs
+- Explicit decision gates and measurable success criteria
+- Clear ownership boundaries with other agents
+
+## Speech Examples
+
+```
+📣 Mia: "Let's pick one beachhead ICP first, then expand."
+📣 Mia: "Message-market fit is weak. The claim and proof don't align yet."
+📣 Mia: "We'll define launch gates now: scale, stop, or pivot by week 1."
+```
+
+## Usage
+
+```
+/gtm             # Activate Mia + display menu
+/gtm strategic   # Strategic tone (direction/trade-off)
+/gtm execution   # Execution tone (owner/timeline/checklist)
+/gtm strategy    # End-to-end GTM blueprint
+/gtm message     # Messaging map refinement
+/gtm launch      # Launch playbook
+/gtm experiment  # GTM experiment backlog
+```
+
+---
+
+## Tone Mode Selection
+
+| Mode | Use Case | Output Style |
+|------|----------|--------------|
+| **Strategic** | "Which direction should we take?" | options/trade-offs/recommendation |
+| **Execution** | "What do we do this week?" | owner/timeline/checklist/KPI cadence |
+
+If not specified, Mia auto-selects mode from the request and states it explicitly.
+
+---
+
+## Trigger Menu
+
+| Trigger | Function |
+|---------|----------|
+| **GTM** | GTM blueprint (ICP + positioning + channel + KPI) |
+| **MSG** | Messaging map (problem/value/proof/CTA by segment) |
+| **LCH** | Launch playbook (pre-launch / launch / post-launch) |
+| **EXP** | Experiment backlog (acquisition/activation tests) |
+| **PKG** | Packaging narrative (plan/price story) |
+| **STM** | Strategic mode response (direction decision support) |
+| **EXM** | Execution mode response (launch runbook style) |
+
+---
+
+## Handoff Flow
+
+```
+🗺️ Marco (market evidence) + 🎤 Rita (VOC)
+                 ↓
+            📣 Mia (GTM design)
+                 ↓
+       📋 Penny / 📡 Tara / 📊 Vicky
+```
+
+---
+
+*Agent*: 📣 Mia (GTM Strategist)
+*Connections*: 🎯 Simon (strategy owner), 🗺️ Marco (market), 🎤 Rita (VOC), 📈 Danny (analysis), 📋 Penny (planning), 📡 Tara (tracking), 📊 Vicky (validation)

package/adapters/claude-code/.claude/commands/handoff.md

@@ -246,6 +246,7 @@ Automatically suggest handoff at work completion:
 | Situation | Suggestion |
 |-----------|-----------|
 | PRD writing complete | "Shall we hand off to 📋 Penny?" |
+| Product direction finalized | "Shall we hand off to 📣 Mia for GTM planning?" |
 | Story writing complete | "Shall we request screen specs from 📐 Hank?" |
 | Screen spec complete | "Shall we hand off to 🔨 Derek?" |
 | Spec has tracking events | "Shall we request review from 📡 Tara?" |
@@ -254,5 +255,5 @@ Automatically suggest handoff at work completion:
 
 ---
 
-*Connected agents*: 🎯 Simon, 📋 Penny, 📐 Hank, 📡 Tara, 🔨 Derek, 🧪 Quinn, 📊 Vicky, 🗓️ Nora
-*Related commands*: `/strategy`, `/plan`, `/market`, `/tracking`, `/validate`, `/retro`
+*Connected agents*: 🎯 Simon, 📣 Mia, 📋 Penny, 📐 Hank, 📡 Tara, 🔨 Derek, 🧪 Quinn, 📊 Vicky, 🗓️ Nora
+*Related commands*: `/strategy`, `/gtm`, `/plan`, `/market`, `/tracking`, `/validate`, `/retro`

package/adapters/claude-code/.claude/commands/market.md

@@ -116,5 +116,5 @@ If `$ARGUMENTS` is `research`:
 ---
 
 *Agent*: 🗺️ Marco (Market Researcher)
-*Connections*: 🎯 Simon (strategic decisions), 📈 Danny (market data validation)
+*Connections*: 🎯 Simon (strategic decisions), 📣 Mia (GTM strategy), 📈 Danny (market data validation)
 *Tools*: WebSearch, WebFetch

package/adapters/claude-code/.claude/commands/start.md

@@ -19,14 +19,14 @@ ARGUMENTS: $ARGUMENTS
 
 ## Execution Steps
 
-### -1. Setup Check (First Run Detection) ⭐
+### -1. Setup + Ambiguity Gate Check ⭐
 
 **First**, check whether the `.context/project.yaml` file exists.
 
 ```
 project.yaml exists?
-├── Yes → Normal flow (proceed to step 0)
-└── No  → Start Setup Wizard
+├── No  → Start Setup Wizard
+└── Yes → Evaluate ambiguity gate before normal flow
 ```
 
 #### Setup Wizard Start
@@ -54,13 +54,10 @@ If `.context/project.yaml` does not exist, treat this as a first run and start t
    - Communication style, work style preferences
    - → Generate `user-context.yaml`
 
-4. **Phase 2: In-depth Project Interview** (🎯 Simon deployed)
-   - Core problem, target customers, market
-   - Solution, differentiators, expected outcomes
-   - Current stage, uncertainties, milestones
-   - What's validated / what needs validation
-   - Research unknown areas together via WebSearch
-   - → Generate `project.yaml`
+4. **Phase 2: Project Skeleton Setup**
+   - Fill minimal project defaults and metadata gate fields
+   - Keep deep interview for ambiguity-gate phase
+   - → Generate initial `project.yaml`
 
 5. **Phase 3: Sensitive Information Guidance** (🎩 Oscar)
    - Suggest creating `.secrets.yaml` template
@@ -76,6 +73,37 @@ Generated files:
 What would you like to start working on?
 ```
 
+#### Ambiguity Gate (mandatory before normal `/start`)
+
+If `.context/project.yaml` exists, read `_meta` and evaluate:
+
+```js
+meta = _meta ?? {};
+needsDeepInterview = meta.needs_deep_interview === true;
+ambiguityScoreRaw = meta.ambiguity_score ?? (needsDeepInterview ? 1 : 0);
+ambiguityScoreParsed = Number(ambiguityScoreRaw);
+ambiguityScore = Number.isFinite(ambiguityScoreParsed) ? ambiguityScoreParsed : 1;
+
+if (needsDeepInterview || ambiguityScore >= 0.6) {
+  // Gate is open: run deep interview first
+} else {
+  // Gate is closed: proceed to step 0
+}
+```
+
+If `_meta` is missing or malformed, treat it as **gate open** (safe default).
+If gate is open, **do not proceed to step 0/session selection yet**.
+Run the deep project interview first (🎯 Simon-led), then update `.context/project.yaml`:
+
+```yaml
+_meta:
+  needs_deep_interview: false
+  ambiguity_score: <updated numeric score>
+  last_interview: "<ISO8601>"
+```
+
+Only after this metadata update should `/start` continue with normal session flow.
+
 ---
 
 ### 0. Load Secret Variables (Required)

package/adapters/claude-code/.claude/commands/strategy.md

@@ -34,7 +34,7 @@ Read `.context/agents/strategist.md` and activate Simon's persona.
 | **HYP** | Formulate hypothesis (Problem → Hypothesis → Validation method) |
 | **PRI** | Prioritization (Impact vs Effort) |
 
-> **Note**: Market analysis → 🗺️ Marco (`/market`). Simon requests market data from Marco.
+> **Note**: Market analysis → 🗺️ Marco (`/market`). GTM planning → 📣 Mia (`/gtm`).
 
 ## Response Format
 

package/adapters/claude-code/CLAUDE.md.hbs

@@ -52,6 +52,7 @@ Oscar is prohibited from answering directly. When delegating, explain the reason
 | `VOC`, `고객/customer`, `피드백/feedback`, `불만/complaint`, `인터뷰/interview` | 🎤 **Rita** | `agents/researcher.md` |
 | `가설/hypothesis`, `전략/strategy`, `PRD`, `우선순위/priority`, `왜/why` | 🎯 **Simon** | `agents/strategist.md` |
 | `시장/market`, `경쟁사/competitor`, `포지셔닝/positioning`, `벤치마크/benchmark` | 🗺️ **Marco** | `agents/market-researcher.md` |
+| `GTM`, `go-to-market`, `런치/launch`, `메시징/messaging`, `채널/channel` | 📣 **Mia** | `agents/gtm-strategist.md` |
 | `스프린트/sprint`, `스토리/story`, `백로그/backlog`, `속도/velocity` | 📋 **Penny** | `agents/planner.md` |
 | `스크린스펙/screen spec`, `핸드오프/handoff`, `릴리스/release`, `개발전달/dev delivery` | 📐 **Hank** | `agents/handoff-specialist.md` |
 | `검증/validation`, `Before/After`, `Guard Rail`, `OKR check` | 📊 **Vicky** | `agents/validator.md` |
@@ -129,13 +130,14 @@ Include Working Memory in the handoff context when switching agents.
 
 ### Agents
 
-#### PO Team (8 + Orchestrator + Ollie/Sage)
+#### PO Team (10 + Orchestrator + Ollie/Sage)
 
 | Command | Agent | Role |
 |---------|-------|------|
 | (default) | 🎩 Oscar | **Orchestrator** — Routing, coordination, proactive alerts |
 | `/strategy` | 🎯 Simon | PRD, hypotheses, priorities |
 | `/market` | 🗺️ Marco | Market research, competitive analysis, positioning |
+| `/gtm` | 📣 Mia | Go-to-market strategy, launch planning, messaging |
 | `/plan` | 📋 Penny | Sprint, stories, backlog |
 | `/handoff` | 📐 Hank | Screen specs, handoff, release readiness |
 | `/validate` | 📊 Vicky | Hypothesis validation, Guard Rail, OKR check |
@@ -171,9 +173,9 @@ Include Working Memory in the handoff context when switching agents.
 ├── .secrets.yaml         # Sensitive data (gitignored)
 ├── WORKFLOW.md           # Workflow guide
 │
-├── agents/               # Agent personas (14 + TEMPLATE.md)
+├── agents/               # Agent personas (15 + TEMPLATE.md)
 ├── oscar/workflows/      # Oscar extension modules (conditionally loaded)
-├── templates/            # Reusable templates (11)
+├── templates/            # Reusable templates (12)
 ├── metrics/              # Metrics data
 ├── daily/                # Daily work logs (YYYY-MM-DD.yaml)
 │

package/bin/cli.mjs

@@ -13,7 +13,7 @@ const USAGE = `
 
   Commands:
     init [dir]      Scaffold + interactive setup + hydration (default)
-    hydrate [dir]   Re-hydrate .hbs templates from existing project.yaml
+    hydrate [dir]   Sync latest scaffold templates + re-hydrate from project.yaml
     doctor [dir]    Check installation health
     help            Show this help
 
@@ -26,6 +26,7 @@ const USAGE = `
   Examples:
     npx popilot init my-project
     npx popilot hydrate
+    npx popilot hydrate --force
     npx popilot doctor
     npx popilot my-project          # same as: popilot init my-project
 `;
@@ -65,7 +66,7 @@ async function main() {
       await cmdInit(targetDir, { skipSpecSite, force, platform });
       break;
     case 'hydrate':
-      await cmdHydrate(targetDir, { skipSpecSite, platform });
+      await cmdHydrate(targetDir, { skipSpecSite, platform, force });
       break;
     case 'doctor':
       await cmdDoctor(targetDir, { skipSpecSite, platform });
@@ -183,12 +184,33 @@ async function cmdInit(targetDir, { skipSpecSite, force, platform }) {
 
 // ── hydrate ─────────────────────────────────────────────
 
-async function cmdHydrate(targetDir, { skipSpecSite, platform }) {
+async function cmdHydrate(targetDir, { skipSpecSite, platform, force }) {
   console.log();
   console.log('  🚀 Popilot — Re-hydrating templates');
   console.log('  ══════════════════════════════════════');
   console.log();
 
+  // 0. Sync scaffold updates before hydration.
+  // - default: only add missing files (safe upgrade path)
+  // - --force: overwrite existing scaffold files as well
+  console.log(`  🔄 Syncing latest scaffold (${force ? 'overwrite enabled' : 'missing files only'})...`);
+  const { copied, overwritten, appends } = await copyScaffold(targetDir, {
+    skipSpecSite,
+    overwriteExisting: force,
+    platform,
+  });
+
+  for (const { file, content } of appends) {
+    const filePath = resolve(targetDir, file);
+    await appendToFile(filePath, content);
+  }
+
+  console.log(`     +${copied.length} files added`);
+  if (overwritten.length > 0) {
+    console.log(`     ~${overwritten.length} files overwritten (--force)`);
+  }
+
+  // 1. Hydrate
   const { hydrated, domains } = await hydrate(targetDir, { skipSpecSite, platform });
   for (const f of hydrated) {
     console.log(`     ${f} ✅`);

package/lib/hydrate.mjs

@@ -75,6 +75,7 @@ export async function hydrate(targetDir, opts = {}) {
     'orchestrator.md.hbs': 'orchestrator',
     'strategist.md.hbs': 'strategist',
     'market-researcher.md.hbs': 'market-researcher',
+    'gtm-strategist.md.hbs': 'gtm-strategist',
     'planner.md.hbs': 'planner',
     'handoff-specialist.md.hbs': 'handoff-specialist',
     'validator.md.hbs': 'validator',

package/lib/setup-wizard.mjs

@@ -2,7 +2,7 @@
  * Interactive setup wizard — terminal interview → project.yaml + related files.
  *
  * Collects only essential config. AI deep interview is deferred to Claude Code
- * via `_meta.needs_deep_interview: true`.
+ * via metadata gate fields (`_meta.needs_deep_interview`, `_meta.ambiguity_score`).
  */
 
 import { readdir, readFile, writeFile, mkdir } from 'node:fs/promises';
@@ -16,6 +16,7 @@ import { parse as parseYaml, stringify as stringifyYaml } from './yaml-lite.mjs'
  * @param {string} targetDir - Project root (scaffold already copied)
  * @param {object} [opts]
  * @param {import('node:readline/promises').Interface} [opts.rl] - Inject readline for testing
+ * @param {string|null} [opts.platform] - Optional adapter/platform name
  * @returns {Promise<void>}
  */
 export async function runSetupWizard(targetDir, opts = {}) {
@@ -232,14 +233,76 @@ async function collectObjectList(rl, question) {
   return items;
 }
 
+// ── Ambiguity score ─────────────────────────────────────
+
+/**
+ * Calculate ambiguity score from project.yaml content.
+ * Returns 0.0 (clear) to 1.0 (high ambiguity).
+ *
+ * Note: fresh setup defaults are usually high ambiguity because deep-interview
+ * fields are empty, but this may be below 1.0 if users fill optional basics
+ * (for example `project.tagline`) during setup.
+ *
+ * @param {object} yaml
+ * @returns {number}
+ */
+export function calculateAmbiguityScore(yaml) {
+  const root = (yaml && typeof yaml === 'object') ? yaml : {};
+
+  const checks = [
+    // Basic clarity
+    { weight: 0.1, filled: !!root.project?.tagline },
+
+    // Problem & market
+    { weight: 0.15, filled: !!root.problem?.core },
+    { weight: 0.1, filled: !!root.problem?.target },
+    { weight: 0.05, filled: root.problem?.alternatives?.length > 0 },
+    { weight: 0.05, filled: !!root.problem?.timing },
+
+    // Solution
+    { weight: 0.15, filled: !!root.solution?.approach },
+    { weight: 0.08, filled: !!root.solution?.differentiation },
+    { weight: 0.05, filled: root.solution?.outcome?.length > 0 },
+
+    // Current state
+    { weight: 0.1, filled: !!root.current_state?.stage },
+    { weight: 0.07, filled: !!root.current_state?.focus },
+    { weight: 0.05, filled: !!root.current_state?.next_milestone },
+
+    // Validation
+    { weight: 0.05, filled: root.validation?.confirmed?.length > 0 },
+  ];
+
+  let totalWeight = 0;
+  let filledWeight = 0;
+
+  for (const { weight, filled } of checks) {
+    totalWeight += weight;
+    if (filled) filledWeight += weight;
+  }
+
+  return Math.round((1 - filledWeight / totalWeight) * 100) / 100;
+}
+
 // ── project.yaml builder ────────────────────────────────
 
+export const ALL_INTEGRATION_PROVIDERS = [
+  'ga4',
+  'mixpanel',
+  'notion',
+  'linear',
+  'channel_io',
+  'intercom',
+  'prod_db',
+  'notebooklm',
+  'corti',
+];
+
 function buildProjectYaml({ projectName, tagline, projectType, domains, devScope, integrations, platform }) {
   // Build the full integrations block with all known providers
-  const allProviders = ['ga4', 'mixpanel', 'notion', 'linear', 'channel_io', 'intercom', 'prod_db', 'notebooklm', 'corti'];
   const integrationsBlock = {};
 
-  for (const id of allProviders) {
+  for (const id of ALL_INTEGRATION_PROVIDERS) {
     if (integrations[id]) {
       integrationsBlock[id] = integrations[id];
     } else {
@@ -247,7 +310,7 @@ function buildProjectYaml({ projectName, tagline, projectType, domains, devScope
     }
   }
 
-  return {
+  const yaml = {
     project: {
       name: projectName,
       tagline: tagline || '',
@@ -298,11 +361,16 @@ function buildProjectYaml({ projectName, tagline, projectType, domains, devScope
       created_at: new Date().toISOString(),
       created_by: 'popilot init',
       needs_deep_interview: true,
+      ambiguity_score: 0,
       last_interview: null,
       version: '1.0.0',
       ...(platform ? { platform } : {}),
     },
   };
+
+  yaml._meta.ambiguity_score = calculateAmbiguityScore(yaml);
+
+  return yaml;
 }
 
 // ── Helpers ─────────────────────────────────────────────

package/package.json

@@ -1,17 +1,20 @@
 {
   "name": "popilot",
-  "version": "0.2.0",
+  "version": "0.3.0",
   "description": "Multi-agent PO/PM system scaffold for Claude Code",
   "type": "module",
   "bin": {
-    "popilot": "./bin/cli.mjs"
+    "popilot": "bin/cli.mjs"
   },
   "scripts": {
+    "lint": "node scripts/lint.mjs",
+    "typecheck": "node scripts/typecheck.mjs",
     "test": "node --test test/*.test.mjs"
   },
   "files": [
     "bin/",
     "lib/",
+    "scripts/",
     "scaffold/",
     "adapters/",
     "README.md"

package/scaffold/.context/WORKFLOW.md.hbs

@@ -56,10 +56,10 @@ The Oscar system and service codebase are in **separate Git repositories**:
                        🎩 Oscar
                   (Routing & Coordination)
                          │
-       ┌─────────────────┼─────────────────┐
-       │                 │                 │
-       ▼                 ▼                 ▼
-   🎯 Simon ◄──────► 📈 Danny ◄──────► 🎤 Rita
+       ┌─────────────────┼─────────────────┬──────────┐
+       │                 │                 │          │
+       ▼                 ▼                 ▼          ▼
+   🎯 Simon ◄──────► 📈 Danny ◄──────► 🎤 Rita      📣 Mia
        │                 │
        ▼                 │
    📋 Penny              │
@@ -74,6 +74,8 @@ The Oscar system and service codebase are in **separate Git repositories**:
 | From | To | Situation | Oscar's Role |
 |------|-----|-----------|-------------|
 | 🎯 Simon | 📋 Penny | PRD completed | Handoff check |
+| 🎯 Simon | 📣 Mia | Product direction finalized | Trigger GTM blueprint |
+| 📣 Mia | 📋 Penny | GTM plan finalized | Convert GTM scope to sprint stories |
 | 📋 Penny | 📊 Vicky | Development completed | Confirm validation readiness |
 | 📊 Vicky | 🎯 Simon | Validation completed | Prompt learning summary |
 | 📊 Vicky | 📈 Danny | Anomaly detected | Request deep analysis |
@@ -105,6 +107,10 @@ Level 1: Sprint PRD (Sprint level)
    └─→ One Question, hypothesis, OMTM
    └─→ Owner: 🎯 Simon / Template: templates/prd.md
 
+Level 1.5: GTM Plan (Launch level, optional but recommended)
+   └─→ ICP, messaging, channel, launch KPI/gates
+   └─→ Owner: 📣 Mia / Template: templates/gtm-plan.md
+
 Level 2: Epic Spec (Epic level) ⭐ Core
    └─→ WHY, WHAT, HOW (overview), EDGE
    └─→ Owner: PO (🎯 Simon support) / Template: templates/epic-spec.md
@@ -268,6 +274,7 @@ Turso (LibSQL) based team retrospective board. Route: `/retro`.
 | Template | Owner | File | Order |
 |----------|-------|------|-------|
 | Sprint PRD | 🎯 Simon | `templates/prd.md` | 1 |
+| GTM Plan | 📣 Mia | `templates/gtm-plan.md` | 1.5 |
 | Epic Spec | PO | `templates/epic-spec.md` | 2 |
 
 ### Handoff (📋 Penny-led)

package/scaffold/.context/agents/TEMPLATE.md

@@ -112,4 +112,4 @@ read_only: true | false                # Whether this agent can modify files
 ---
 
 *Version*: v1.0
-*Applies to*: All 14 agents in the Oscar system
+*Applies to*: All 15 agents in the Oscar system