spec-driven-feedback

name	spec-driven-feedback
description	Retrospective feedback system with Execute-Verify-Gate structural anti-skip enforcement for DevForgeAI operations. Captures feedback after /dev, /qa, /release, sprint planning, or manual triggers. Handles 6 feedback types: conversation, summary, metrics, checklist, ai_analysis, and triage. Prevents token optimization bias through lean orchestration, per-phase reference loading, checkpoint persistence, and binary CLI gate enforcement. Use when feedback needs to be captured, when the user mentions retrospectives, lessons learned, workflow improvements, process feedback, or wants to review what went well or poorly. Also handles AI architectural analysis, recommendation triage, feedback search, export/import, and configuration management.
allowed-tools	["Read","Write","Edit","Glob","Grep","AskUserQuestion","Task","Skill","Bash(python3:*)"]
model	claude-opus-4-7
effort	High
version	1.0.0
topics	feedback, retrospective, anti-skip, spec-driven, hooks, triage, ai-analysis

Spec-Driven Feedback

Capture retrospective feedback from development workflows to improve processes, identify patterns, and enable continuous improvement — with structural anti-skip enforcement.

If ambiguous or conflicts detected: HALT and use AskUserQuestion

---

Execution Model

This skill expands inline. After invocation, execute Phase 00 Initialization immediately. Do not wait passively, ask permission, or offer execution options.

Self-Check (if ANY box is true = VIOLATION):

• Stopping to ask about token budget
• Stopping to offer execution options
• Waiting passively for user to say "go"
• Asking "should I execute this?"
• Skipping a phase because it "seems simple"
• Combining multiple phases into one
• Summarizing instead of loading a reference file
• Skipping verification because "I already wrote the file"

IF any box checked: EXECUTION MODEL VIOLATION. Go directly to Phase 00 Initialization now.

---

Anti-Skip Enforcement Contract

This skill enforces 4 independent anti-skip layers. ALL FOUR must fail for a step to be skipped:

1. Per-phase reference loading - Each phase loads its reference files fresh via Read(). NOT consolidated. Prevents "already covered" rationalization.
2. Binary CLI gates - devforgeai-validate phase-check/phase-complete at phase boundaries. Cannot be forged by LLM.
3. Checkpoint-based state tracking - Phase completion verified by checking checkpoint JSON data keys and current_phase field.
4. Artifact verification - Checkpoint JSON existence checked via Glob(), feedback files verified on disk, index entries verified via Grep().

Execute-Verify-Record Pattern: Every mandatory step in every phase file has three parts:

• EXECUTE: The exact action to perform (Read, Write, AskUserQuestion, Task, Grep, Glob)