openspec-reflect

OpenSpec Reflect

Pre-gate self-check skill. Verifies acceptance criteria, detects scope drift, and ensures philosophy alignment before human review.

Workflow: Check → Assess → Report

flowchart LR
    A["Read context"] --> B["Check criteria"]
    B --> C["Detect drift"]
    C --> D["Check philosophy"]
    D --> E["Self-assess"]
    E --> F["Generate report"]

    classDef action fill:#E1BEE7,stroke:#7B1FA2,color:#000
    class A,B,C,D,E,F action

Critical: Run before human gates. Catches drift BEFORE human reviews.

Commands

reflect

Run comprehensive pre-gate self-check.

Input: $ARGUMENTS = change-id

Workflow:

1. Load context:

   • Read openspec/changes/{change-id}/proposal.md for acceptance criteria
   • Read openspec/changes/{change-id}/tasks.md for progress
   • Read openspec/changes/{change-id}/test.md for test strategy (if exists)
   • Read openspec/changes/{change-id}/specs/*.md for requirements
   • Read openspec/project.md for Execution Philosophy

2. Pre-gate check (scale/maintenance mode only):

   • Check if test.md exists
   • If missing in scale/maintenance mode: flag ⚠️ MISSING_TEST_MD deviation
   • In garage mode: test.md optional (note if missing, don't block)

3. Check acceptance criteria:

   • Extract criteria from proposal.md (Success Criteria section)
   • For each criterion: verify met/unmet/partial
   • Record evidence for each status

4. Detect scope drift:

   • Extract "Affected files" from proposal.md
   • Run git status and git diff --stat to find actual changes
   • Calculate deviation percentage: (actual - proposed) / proposed * 100
   • Flag if >20% deviation

5. Check philosophy alignment:

   • Read mode from Execution Philosophy
   • Review implementation for anti-patterns of current mode
   • Flag any violations

6. Self-assessment:

   • Answer: "Am I solving the right problem?"
   • Evaluate: Is work aligned with proposal's Problem statement?