[auto] [Issue #61] Guided onboarding for charter init

[stackbilt-admin]

Autonomous Task

Task ID: eeb9f4ad-5ea4-4f54-a565-5ffef15a79bd
Authority: auto_safe
Exit code: 3

Task Prompt

Problem

charter init currently does minimal scaffolding — creates a .ai/ directory with basic files but provides no guidance. New users don't know which ADF modules to create, how governance policies work, or what the manifest router does.

What to Fix

1. Read the current charter init implementation (likely in packages/cli/src/commands/init.ts or similar).

2. After scaffolding the .ai/ directory, print a guided welcome message:

   ✓ Created .ai/ directory
   
   Your AI governance is ready. Here's what was created:
   
     .ai/manifest.adf    — Module router (controls what loads when)
     .ai/core.adf         — Universal rules (always loaded)
   
   Next steps:
     1. Edit .ai/core.adf to add your project's constraints
     2. Run 'charter doctor' to validate your setup
     3. Run 'charter bundle' to see the assembled context
   
   Docs: https://github.com/Stackbilt-dev/charter
   

3. If the .ai/ directory already exists, show a different message:

   .ai/ directory already exists. Run 'charter doctor' to check for issues.
   

4. Consider adding a --guided flag that asks interactive questions:

   • "What's your primary language?" (TypeScript, Python, Go, etc.)
   • "Do you use Cloudflare Workers?" (adds Workers-specific rules to core.adf)
   • "Do you want auth governance?" (creates auth.adf)
     But ONLY if the flag is passed — default charter init should be non-interactive.

Constraints

• Do NOT change the core ADF format
• Do NOT add new dependencies for the interactive mode
• Keep the default (no flags) non-interactive
• The welcome message should be genuinely helpful, not marketing

Commit your work. TASK_COMPLETE when done.

Result Summary

[max_turns_exceeded] Task ran out of turns (26 used, $1.23). Increase max_turns or simplify the task.

---

Generated by AEGIS task runner. Review before merging.