Skip to content

D1: prefer eligible empty candidates + report eligible vs listed (#172) #173

Description

@xenodeve

Parent

Part of #172

What to build

Improve the existing case-study (and taxonomy if needed) capped generator loop so each run prefers projects that still need work, and dry-run/apply responses distinguish eligible candidates from merely scanned/listed published GitHub projects. Behavior of generation gates (#160 empty+auto, human skip) stays the same — this slice is ordering + reporting so operators and cron make measurable progress without misreading candidates:46.

Acceptance criteria

  • Case-study dry-run reports both listed/scanned and eligible (would attempt LLM) counts
  • Eligible empty+auto projects are preferred before already-complete rows in the capped loop
  • Unit/controller tests cover the reporting and preference behavior at the public service/controller seam
  • Default per-run cap remains 1; no Vercel timeout regression path introduced

Blocked by

None - can start immediately

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or requestready-for-agentFully specified, ready for an AFK agent

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions