SKILL.md 2.1 KB


name: aidop-auto-verify-fix-loop

description: Enforces automatic verify-and-fix loop after code changes in Ai-DOP (Admin.NET + Web). Use when implementing any change, especially tenant, RBAC, menu routing, and kanban data integration.

Ai-DOP Auto Verify Fix Loop

Purpose

After any implementation, run verification immediately, fix failures, and re-run checks until pass or explicit external blocker.

Quick Workflow

  1. Implement requested changes.
  2. Choose validation set by changed area.
  3. Run checks and collect errors.
  4. Fix root cause (not symptom).
  5. Re-run same checks.
  6. Report with: changed files, checks run, results, remaining blockers.

Validation Matrix

Backend C# changes

  • Build target project first:
    • references/Admin.NET/server/Admin.NET.Core: dotnet build --framework net8.0
    • references/Admin.NET/server/Admin.NET.Web.Entry: dotnet build --framework net8.0
  • If API behavior changed, start/restart backend and run endpoint smoke tests.

Frontend Web changes

  • In references/Admin.NET/Web:
    • npm run build
  • If route/menu/request logic changed, also verify:
    • page opens without blank screen
    • /api proxy path works
    • key endpoints return expected status

Tenant / RBAC / Menu changes (mandatory extra checks)

  • Smoke check at least:
    • /api/sysTenant/page
    • /api/sysRole/page
    • /api/sysMenu/list
  • For tenant-switch related changes, verify sysTenant:changeTenant permission path.

AiDOP Kanban / diagnosis changes

  • Verify related endpoints and one page-load smoke.
  • Confirm no new static fallback masks backend failures.

Command Defaults

  • Use npm if pnpm is unavailable.
  • Prefer targeted builds over full-solution build for faster loop.
  • Keep long-running dev servers in background and poll health.

Stop Conditions

  • Stop only when:
    • checks pass, or
    • blocked by external dependency (network/service/credential) with explicit blocker note.

Output Template

Use this final structure:

  • Changes: key files and what changed.
  • Validation: commands/endpoints checked.
  • Result: pass/fail per item.
  • Blockers/Risks: anything user must handle.