Menü
Run correctness, quality, security, consistency, and AI-output gates for Codex-native migration work.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Basierend auf der SOC-Berufsklassifikation
Improve, audit, simplify, rewrite, or migrate prompts, tool descriptions, agent instructions, and prompt stacks for GPT-5.6 Sol or the GPT-5.6 family. Use for outcome-first prompt design, autonomy boundaries, tool routing, PTC, grounding, verbosity, reasoning effort, and prompt evals.
Apply a deep, design-minded engineering workflow for complex work that needs careful planning, iteration, and simplification.
Use when preparing or running a Claude CLI agentic engineering review through claude -p for repository audits, architecture analysis, system design, security review, large refactors, specs, RFCs, or evidence-grounded long-form engineering analysis.
Use when preparing or running a ZCode GLM-5.2 agentic builder workflow through zcode --prompt for fast implementation, iterative code generation, focused fixes, and validation on an existing repository.
Apply adversarial opposite-analysis to plans, specs, architecture, code changes, and claims. Use when the user asks for adversarial review, opposing analysis, contrarian review, red-team reasoning, or Z.ai and MiniMax cross-checks through the Ralph MCP router.
Review and adjudicate Bugbot, Cursor, Seer, and similar automated PR feedback with local evidence before accepting, fixing, or dismissing findings.
| name | gates |
| description | Run correctness, quality, security, consistency, and AI-output gates for Codex-native migration work. |
Verify that generated work is correct, safe, consistent with repo policy, and free of obvious AI-output quality issues before a checkpoint is marked PASS.
Run gates in this order: correctness first, then quality, security, consistency, and AI-output quality. Correctness means the smallest parse, compile, or unit command that proves the artifact is structurally valid. Quality covers lint, format, types, naming, and local style. Security covers secret scans, unsafe config, and RED-content checks. Consistency compares the work against AGENTS.md, migration scope, route policy, and checkpoint requirements.
Use slop-guard as a strong gate for generated Markdown, docs, PR text, checkpoint prose, and user-facing summaries.
Preferred routes when available:
The preferred route is the slop-guard MCP with check_slop or check_slop_file. The local fallback is uvx --from slop-guard sg <file>. Use a plain sg <file> command only after confirming that sg is actually the slop-guard CLI in the active environment.
If slop-guard reports a failing score or moderate/heavy findings, rewrite the affected prose, remove formulaic phrasing, and rerun the gate.
If slop-guard is not installed or not configured, do a manual prose review and record the degraded route. Future setup should make this gate executable and blocking.
Project work must have no literal API keys, no direct Z.ai or MiniMax model_provider entries, and no private vault data. Every required skill must have a SKILL.md with name and description frontmatter. The previous checkpoint must be PASS before the next phase starts. The orchestrator must enforce slop-guard as the strong prose-quality gate.
Security hardening gates must include content-sensitive fixtures for API keys, JWTs, private keys, seed phrases, wallet material, OAuth tokens, database URLs, .env references, and customer-sensitive markers. RED must not persist to vault/runtime reports and must not externalize to MCPs. Coding model evals must keep sensitive_externalization_incidents at zero.
A phase can be PASS only when all blocking gates pass or have a documented safe fallback.