Teach the backlog groomer to generate adoption and credibility issues, not just infrastructure #127
Description
Goal
Update the backlog groomer to consider external adoption signals (GitHub stars, README quality, quickstart friction, demo availability, public proof of work) alongside its current internal inputs (CODEBASE.md known issues, risk flags, task metrics). The groomer currently only generates infrastructure and reliability tasks because its inputs are exclusively internal. This creates a structural bias where the system endlessly polishes its own plumbing while never producing work that would attract users.
Success Criteria
- Groomer prompt includes adoption context: current star count, README length/structure assessment, quickstart friction level, demo availability
- At least 1 of every 5 generated issues should target adoption, credibility, or activation improvement
- Groomer considers the product vision and north star's adoption goals, not just operational metrics
- Generated adoption issues have clear, actionable scope (not vague "improve marketing")
Why This Matters
The backlog groomer is the primary source of new issues for the autonomous loop. If it only generates internal tasks, the planner can only select internal tasks, and the system never works on adoption. This is the structural fix that prevents the infinite self-improvement loop.
Outcome Checks
- github_stars