Codex-SEO：面向Codex的26工作流SEO技能套件与确定性审计工具

AgriciDaniel has open-sourced the Codex-SEO project, a full port of Claude-SEO to the Codex platform. It provides 26 professional SEO workflows, 24 TOML agent configurations, MCP integration, and supports natural language commands as well as API-safe execution. This project addresses key pain points in traditional SEO audits and offers a systematic, AI-native solution for high-quality SEO optimization.

Traditional SEO audits are time-consuming and require manual checks of dozens of technical indicators (e.g., page speed, structured data). While AI integration is an industry trend, professional SEO needs systematic methodologies, seamless tool ecosystem integration (Google Search Console, DataForSEO, Firecrawl), and enterprise-grade features like deterministic output and API-safe execution—gaps Codex-SEO aims to fill.

Codex-SEO is a Codex-first suite including 1 orchestrator skill,26 workflows covering the full SEO lifecycle,24 TOML agent configs, and a shared cache. Key workflows are categorized:

• Basic audit (audit, page, technical, content)
• Structured data & performance (schema, performance, images)
• Site architecture & internationalization (sitemap, hreflang, programmatic)
• Local SEO (local, maps)
• Competition analysis (competitor pages, backlinks, cluster)
• Advanced analysis (GEO, visual, SXO, drift, ecommerce)
• Integration (Google APIs, DataForSEO, Firecrawl, image generation, FLOW)

Codex-SEO is deeply adapted to Codex:

• Skill encapsulation: Core orchestrator (skills/seo/SKILL.md) and modular workflow skills (skills/seo-*).
• Agent configs:24 pre-tuned TOML files (e.g., seo-technical.toml for technical SEO, seo-content.toml for content optimization).
• Plugin discovery: .codex-plugin/plugin.json enables auto-detection.
• Deterministic execution: Scripts like run_skill_workflow.py (headless/API) and run_api_smoke_suite.py (API tests) support CI/CD and batch processing.

Installation:

• macOS/Linux: curl -fsSL https://raw.githubusercontent.com/AgriciDaniel/codex-seo/main/install.sh | bash
• Windows: irm https://raw.githubusercontent.com/AgriciDaniel/codex-seo/main/install.ps1 | iex Options include specifying repo, branch/ref (default v1.9.6-codex.1), Codex home dir, skipping Chromium, etc.

Interaction:

• Natural language: e.g., "Audit https://example.com for SEO."
• Command line: e.g., /seo audit <url>, /seo technical <url>, /seo google setup.

Validation:

• Run tests: python -m pytest tests/
• Verify environment: python scripts/verify_environment.py --json
• Demo readiness: python scripts/demo_readiness.py --target https://example.com --live-apis --workflows --json
• Execute workflow: python scripts/run_skill_workflow.py --skill seo-technical https://example.com --json

Data Isolation:

• New storage: ~/.config/codex-seo/ (credentials), ~/.cache/codex-seo/ (cache).
• Migration: Supports fallback to Claude-SEO paths (~/.config/claude-seo/, ~/.cache/claude-seo/) for compatibility, but new data uses Codex paths.
• Output/cache dirs (output/, .seo-cache/) are git-ignored.

Codex-SEO represents an important step in AI-native SEO tool evolution. It lowers the barrier to high-quality SEO audits by encapsulating professional methodologies into reusable AI agent skills. Beyond SEO, it demonstrates a viable path for AI transformation in other professional services (legal, finance, medical) — not just adding AI chat interfaces, but重构 workflows to make AI agents the core execution engine.