Semantic Verification — Development skill for Claude Code
Stage 11: Independent Semantic Verification.
How to install Semantic Verification
Installs to ~/.claude/skills/cdeust-ai-architect-feedback-loop-semantic_verification/SKILL.md
mkdir -p ~/.claude/skills/cdeust-ai-architect-feedback-loop-semantic_verification && curl -fsSL https://raw.githubusercontent.com/cdeust/ai-architect-feedback-loop/HEAD/prompts/semantic_verification.md -o ~/.claude/skills/cdeust-ai-architect-feedback-loop-semantic_verification/SKILL.md Restart Claude Code, or start a new session, for it to be picked up.
What Semantic Verification does
Stage 11: Independent Semantic Verification
Your Role
You are a VERIFICATION AGENT — completely separate from the implementation agent. Your job is to FIND PROBLEMS, not confirm correctness. Assume the implementation has bugs until proven otherwise.
You are reviewing changes to the target product.
Product Architecture
{{ARCHITECTURE_DESCRIPTION}}
Module Dependency Graph (from engine_graph.json)
{{ENGINE_GRAPH}}
Verification Inputs
PRD Specification (What Should Have Been
Alternatives in Development
- Zeroshot — Runs coding agents as a graph: one agent implements, independent agents review, failures go to repair, and not 1.9k ★
- Fusion Opus4.8 — Fusion panel of two independent Opus 4.8 runs, judged by Opus 4.8 (opus4.8-4.8) 468 ★
- Premiere Pro MCP — Adobe Premiere Pro MCP — independent, local-first server for supported workflows 292 ★
Full documentation available on GitHub
View Source RepositoryRelated Skills
Claude Docu Optimizer
Claude Code plugin for optimizing CLAUDE.md and docs/ ecosystem. Token analysis, anti-pattern detection, seman
AgentScope
Research harness for evaluating Codex and Claude Code across single, staged, and parallel agent architectures
Claude Code Research
Independent research on Claude Code internals, Claude Agent SDK, and related tooling.
Craft Research Verify
Verify existing research findings against independent primary sources. Upgrades confidence from 'sources agree
Hatch3r Ask
Answer a free-form question about the user's codebase, read-only. Triages question breadth, fans out one hatch
Dispatching Parallel Agents
Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies
Related Agents
Gfd Verifier
grace-feature-dev independent verifier. Read-only. Runs the card's tests against the real code and performs Se
Otel Observability Engineer
Independent observability reviewer for a service diff — verifies OpenTelemetry instrumentation, RED metrics (R
Duplication Detector
Detects stories with overlapping implementation concerns using hybrid keyword pre-filter and LLM semantic veri