Implement Context Pack Conversion in textualizer

.agent-guidance/agentic-review-checklist.md

@@ -0,0 +1,17 @@
+# Agentic Review Checklist
+
+When reviewing code generated by agents (LLMs), apply the "External Contributor" mindset.
+
+## 1. Safety & Security
+- [ ] **Secrets:** Did the agent hardcode any credentials?
+- [ ] **Deps:** Did it hallucinate any npm packages?
+- [ ] **Sanitization:** Are inputs validated?
+
+## 2. Logic & Correctness
+- [ ] **Loop:** Did the agent create infinite loops?
+- [ ] **Types:** Are types loose (`any`) where they should be strict?
+- [ ] **Tests:** Do the new tests actually test the logic (or just mock everything)?
+
+## 3. Governance
+- [ ] **Plan:** Does `PLAN.md` match the PR content?
+- [ ] **Drift:** Did the agent modify files outside the plan?

.agent-guidance/testing-and-verification.md

@@ -0,0 +1,100 @@
+# Agent Guidance: Testing & Verification
+
+## Quick Reference for AI Agents
+
+When working on this repository, follow these conventions to avoid introducing flaky tests and CI failures.
+
+### Critical Rules
+
+1. **DO NOT add Jest tests for verification-style checks** (structure, feature presence, config schemas)
+2. **DO NOT mock third-party libraries by hand inside node_modules/**
+3. **DO use node-native tests (tsx + node:assert) for verification**
+4. **DO use Jest only for true unit/integration behavior tests**
+
+### When to Use Each Approach
+
+#### Use Jest (`*.test.ts` in test directories)
+- Testing function/class behavior
+- Unit testing with straightforward mocking
+- React component testing
+- Integration tests requiring test database
+
+#### Use Node-Native Verification (`server/scripts/verify-*.ts`)
+- GA feature presence checks (auth, rate limits, policies)
+- Route/middleware configuration verification
+- Config schema validation
+- Any structural check that doesn't need heavy mocking
+
+### Adding a New Verification
+
+1. Create `server/scripts/verify-your-feature.ts`:
+```typescript
+import assert from 'assert/strict';
+
+async function run() {
+  console.log('--- Your Feature Verification ---');
+
+  // Your checks here using assert.equal(), assert.ok(), etc.
+
+  console.log('✅ All checks passed');
+}
+
+run().catch(err => {
+  console.error('Verification failed:', err);
+  process.exit(1);
+});
+```
+
+2. Add to `scripts/verify.ts`:
+```typescript
+{
+  name: 'Your Feature',
+  category: 'ga-feature' | 'security' | 'structure',
+  script: 'server/scripts/verify-your-feature.ts',
+  required: true,
+}
+```
+
+3. Test: `pnpm verify --filter=your-feature`
+
+### CI Golden Path
+
+The CI pipeline runs:
+1. **Bootstrap** - Install deps, validate versions
+2. **Lint** - ESLint + Ruff
+3. **Verify** - Run verification suite (BLOCKS on failure)
+4. **Test** - Jest unit/integration (currently non-blocking for unit)
+5. **Build** - TypeScript compilation
+
+### Common Pitfalls to Avoid
+
+❌ **DON'T**: Add a Jest test that imports server code with ESM issues
+✅ **DO**: Create a node-native verification script using tsx
+
+❌ **DON'T**: Mock entire dependency chains for a simple structure check
+✅ **DO**: Use direct imports and introspection in verification scripts
+
+❌ **DON'T**: Add to the Jest quarantine list without considering alternatives
+✅ **DO**: Ask if this should be a verification script instead
+
+❌ **DON'T**: Create tests that depend on external services in the verification suite
+✅ **DO**: Keep verifications fast and deterministic with mocks/fixtures
+
+### Examples
+
+**Good Verification Script** (server/scripts/verify-route-rate-limit.ts):
+- Direct imports
+- Mock only what's needed
+- Fast execution
+- Clear pass/fail
+- No Jest dependency
+
+**Good Jest Test** (server/tests/utils/encryption.test.ts):
+- Tests behavior of encryption functions
+- Uses Jest's mocking features appropriately
+- Isolated unit test
+- Doesn't fight ESM transforms
+
+### Further Reading
+
+See [TESTING.md](../TESTING.md) for complete documentation.

.agentic-prompts/README.md

@@ -0,0 +1,72 @@
+# Agentic Prompts Directory
+
+Task-specific prompts for AI coding assistants (Claude Code, Codex, Jules, etc.).
+
+## Structure
+
+```
+.agentic-prompts/
+  README.md                    # This file
+  velocity.log                 # Task completion metrics
+  task-[ID]-[name].md         # Active task prompts
+  archived/                    # Completed task prompts
+    YYYY-MM/
+      task-[ID]-[name].md
+```
+
+## Workflow
+
+### 1. Create Task Prompt
+```bash
+cp prompts/claude-code.md .agentic-prompts/task-456-new-feature.md
+```
+
+### 2. Edit with Requirements
+- Add specific requirements
+- Define success criteria
+- List affected systems
+- Note constraints
+
+### 3. Execute with AI Assistant
+- Load prompt into Claude Code/Codex/etc.
+- Let AI deliver complete implementation
+- Verify quality gates pass
+
+### 4. Create PR
+```bash
+gh pr create --fill
+```
+
+### 5. Archive When Merged
+```bash
+mkdir -p archived/$(date +%Y-%m)
+mv task-456-new-feature.md archived/$(date +%Y-%m)/
+```
+
+## Templates
+
+All agent prompt templates are in `../prompts/`:
+- `claude-code.md` - For complex architectural work
+- `codex.md` - For deterministic critical code
+- `jules-gemini.md` - For cross-file refactoring
+- `cursor-warp.md` - For devloop integration
+- `summit-intelgraph.md` - For multi-service work
+- `ci-cd.md` - For pipeline changes
+
+See `../prompts/meta-router.md` for agent selection guidance.
+
+## Metrics
+
+Target performance:
+- 3-5 tasks per day
+- <2 hours to PR
+- >95% CI pass rate
+- >90% first-time merge rate
+
+## Best Practices
+
+1. **One prompt per task** - Keep focused
+2. **Archive when done** - Keep directory clean
+3. **Log completion** - Track velocity
+4. **Review weekly** - Identify patterns
+5. **Improve templates** - Evolve based on learnings