⚡ Hermes SuperCode Skills
May 28, 2026 · View on GitHub
13 production-grade skill modules for OpenAI Codex · AGENTS.md native · MIT License
Claude Code version: hermes-supercode-skills
English
What is this?
Hermes Codex Edition adapts 13 specialized skill modules for OpenAI Codex via AGENTS.md. Each module activates automatically when a task matches its domain — no manual invocation needed.
Skills cover: database optimization, authentication systems, observability, infrastructure, security, deployments, debugging, API design, data pipelines, prompt engineering, web scraping, prediction markets, and multi-agent orchestration.
Skills
| # | Skill | Domain | Trigger examples |
|---|---|---|---|
| 1 | infra-automation | DNS, SSL, Cloudflare, CDN | "Add an A record for api.example.com" |
| 2 | prediction-alpha | Prediction market analysis | "Analyze this Polymarket market" |
| 3 | ghost-scraper | Ethical web scraping | "Scrape product data from this URL" |
| 4 | mcp-conductor | Multi-agent orchestration | "Coordinate 5 agents on this task" |
| 5 | pipeline-architect | ETL/ELT, streaming, data warehouse | "Build a Kafka-to-ClickHouse pipeline" |
| 6 | security-sentinel | Security audits, vuln scanning | "Audit example.com security posture" |
| 7 | deploy-ninja | Zero-downtime deployments | "Deploy v2.3.4 with canary strategy" |
| 8 | quantum-debugger | Race conditions, memory leaks | "Production service intermittently segfaults" |
| 9 | api-sculptor | REST/GraphQL/gRPC design | "Design an Order API for e-commerce" |
| 10 | prompt-forge | Prompt engineering | "Write a system prompt for a support agent" |
| 11 | db-whisperer | Query optimization, migrations | "This Postgres query is timing out" |
| 12 | auth-architect | OAuth2, JWT, RBAC, SSO, MFA | "Implement OAuth2 with refresh token rotation" |
| 13 | obs-guardian | OpenTelemetry, Prometheus, SLOs | "Add distributed tracing to my microservices" |
Installation
Option A: Use as AGENTS.md in your project
# Copy AGENTS.md to your project root
curl -sO https://raw.githubusercontent.com/mturac/hermes-supercode-skills-codex/main/AGENTS.md
# Codex will automatically pick up the skill definitions
Option B: Full clone with skill reference files
git clone https://github.com/mturac/hermes-supercode-skills-codex.git
# Place AGENTS.md in your project root
cp hermes-supercode-skills-codex/AGENTS.md ./AGENTS.md
Demo
Place AGENTS.md in your project root:
curl -sO https://raw.githubusercontent.com/mturac/hermes-supercode-skills-codex/main/AGENTS.md
Codex now has 13 specialized skill modules. Example session:
User: This Postgres query is taking 8 seconds
Codex: [db-whisperer activated] Running EXPLAIN ANALYZE...
→ Missing index on orders.user_id + created_at
→ Proposed fix: CREATE INDEX CONCURRENTLY idx_orders_user_created
ON orders(user_id, created_at DESC);
→ Estimated improvement: 8s → 12ms (99.8% reduction)
How it works
Codex reads AGENTS.md at the start of every session. The 13 skill definitions are loaded as context — when your task matches a skill's domain, Codex activates that skill's Recon → Plan → Execute → Verify workflow automatically.
Design Principles
- Recon → Plan → Execute → Verify — no skill jumps straight to action
- Structured JSON output — every skill produces machine-parseable results
- Tiered safety rails — 🔴 Red (never) / 🟡 Yellow (confirm first) / 🟢 Green (safe)
- Self-contained — copy
AGENTS.mdinto any project, no dependencies
中文
这是什么？
Hermes Codex 版本通过 AGENTS.md 为 OpenAI Codex 适配了 13 个专业技能模块。每个模块在任务匹配其领域时自动激活，无需手动调用。
技能覆盖：数据库优化、身份认证、可观测性、基础设施、安全、部署、调试、API 设计、数据管道、提示词工程、网页抓取、预测市场、多智能体编排。
技能列表
| # | 技能 | 领域 | 触发示例 |
|---|---|---|---|
| 1 | infra-automation | DNS、SSL、Cloudflare、CDN | "为 api.example.com 添加 A 记录" |
| 2 | prediction-alpha | 预测市场分析 | "分析这个 Polymarket 市场" |
| 3 | ghost-scraper | 合规网页抓取 | "从这个 URL 抓取产品数据" |
| 4 | mcp-conductor | 多智能体编排 | "协调 5 个代理完成此任务" |
| 5 | pipeline-architect | ETL/ELT、流处理、数仓 | "构建 Kafka 到 ClickHouse 的管道" |
| 6 | security-sentinel | 安全审计、漏洞扫描 | "审计 example.com 安全状况" |
| 7 | deploy-ninja | 零停机部署 | "使用金丝雀策略部署 v2.3.4" |
| 8 | quantum-debugger | 竞态条件、内存泄漏 | "生产服务间歇性崩溃" |
| 9 | api-sculptor | REST/GraphQL/gRPC 设计 | "为电商设计订单 API" |
| 10 | prompt-forge | 提示词工程 | "为支持代理编写系统提示词" |
| 11 | db-whisperer | 查询优化、迁移 | "这个 Postgres 查询超时了" |
| 12 | auth-architect | OAuth2、JWT、RBAC、SSO、MFA | "实现带刷新令牌轮换的 OAuth2" |
| 13 | obs-guardian | OpenTelemetry、Prometheus、SLO | "为微服务添加分布式追踪" |
安装
# 将 AGENTS.md 复制到您的项目根目录
curl -sO https://raw.githubusercontent.com/mturac/hermes-supercode-skills-codex/main/AGENTS.md
Codex 在每次会话开始时自动读取 AGENTS.md，13 个技能定义作为上下文加载。
Türkçe
Bu nedir?
Hermes Codex Edition, OpenAI Codex için AGENTS.md üzerinden 13 uzmanlaşmış skill modülünü aktive eder. Her modül, görev kendi alanıyla eşleştiğinde otomatik olarak devreye girer — manuel çağrı gerekmez.
Skill'ler şunları kapsar: veritabanı optimizasyonu, kimlik doğrulama, gözlemlenebilirlik, altyapı, güvenlik, deployment, hata ayıklama, API tasarımı, veri pipeline'ları, prompt mühendisliği, web scraping, tahmin piyasaları, çok ajanlı orkestrasyon.
Skill Listesi
| # | Skill | Alan | Tetikleyici örnekler |
|---|---|---|---|
| 1 | infra-automation | DNS, SSL, Cloudflare, CDN | "api.example.com için A kaydı ekle" |
| 2 | prediction-alpha | Tahmin piyasası analizi | "Bu Polymarket piyasasını analiz et" |
| 3 | ghost-scraper | Etik web scraping | "Bu URL'den ürün verisi çek" |
| 4 | mcp-conductor | Çok ajanlı orkestrasyon | "5 ajanı bu görevde koordine et" |
| 5 | pipeline-architect | ETL/ELT, streaming, veri ambarı | "Kafka'dan ClickHouse'a pipeline kur" |
| 6 | security-sentinel | Güvenlik denetimi, zafiyet tarama | "example.com güvenlik durumunu denetle" |
| 7 | deploy-ninja | Sıfır-kesinti deployment | "v2.3.4'ü canary stratejisiyle deploy et" |
| 8 | quantum-debugger | Race condition, bellek sızıntısı | "Production servis aralıklı çöküyor" |
| 9 | api-sculptor | REST/GraphQL/gRPC tasarımı | "E-ticaret için Sipariş API'si tasarla" |
| 10 | prompt-forge | Prompt mühendisliği | "Destek ajanı için sistem promptu yaz" |
| 11 | db-whisperer | Sorgu optimizasyonu, migrasyonlar | "Bu Postgres sorgusu zaman aşımına uğruyor" |
| 12 | auth-architect | OAuth2, JWT, RBAC, SSO, MFA | "Refresh token rotasyonlu OAuth2 uygula" |
| 13 | obs-guardian | OpenTelemetry, Prometheus, SLO'lar | "Mikroservislerime dağıtık izleme ekle" |
Kurulum
# AGENTS.md'yi proje köküne kopyala
curl -sO https://raw.githubusercontent.com/mturac/hermes-supercode-skills-codex/main/AGENTS.md
Codex her oturumun başında AGENTS.md dosyasını okur. 13 skill tanımı bağlam olarak yüklenir ve göreviniz bir skill'in alanıyla eşleştiğinde otomatik olarak Keşif → Planlama → Uygulama → Doğrulama iş akışı devreye girer.
Repo Structure
hermes-supercode-skills-codex/
├── AGENTS.md ← Drop this in your project root
├── README.md
├── LICENSE
├── skills/
│ ├── db-whisperer/SKILL.md
│ ├── auth-architect/SKILL.md
│ ├── obs-guardian/SKILL.md
│ ├── infra-automation/SKILL.md
│ ├── security-sentinel/SKILL.md
│ ├── deploy-ninja/SKILL.md
│ ├── quantum-debugger/SKILL.md
│ ├── api-sculptor/SKILL.md
│ ├── pipeline-architect/SKILL.md
│ ├── mcp-conductor/SKILL.md
│ ├── ghost-scraper/SKILL.md
│ ├── prediction-alpha/SKILL.md
│ └── prompt-forge/SKILL.md
└── docs/DESIGN_DECISIONS.md
Related
- Claude Code version: hermes-supercode-skills — uses Claude Code's
Skilltool mechanism - Codex automations: awesome-codex-automations
License
MIT — use it, fork it, extend it.