The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and analyze external, untrusted content from the web which could contain malicious instructions designed to influence the agent's behavior during the audit process.
Ingestion points: External robots.txt, sitemap.xml, and raw HTML content (including <title>, <meta> tags, and JSON-LD structured data) are fetched and processed by the skill (Sections 4.1, 4.2, 4.5, 4.6).
Boundary markers: The skill mandates a 'Script-first, AI interpretation second' architecture (Section 3), requiring deterministic script outputs to be saved to JSON evidence files before any AI synthesis occurs (Section 9).
Capability inventory: The skill uses Bash for shell execution, curl for network requests, and Write/Edit for file system modifications (Frontmatter, Section 8).
Sanitization: External data is processed via specific parsers (e.g., robotspec, urllib.robotparser) and validated against schemas (Schema Markup Validator, XML schema validation) before being presented to the agent (Section 4.1, 4.6).
[EXTERNAL_DOWNLOADS]: The skill downloads and executes several well-known and trusted third-party auditing tools and interacts with official service APIs.
Evidence: Uses npx to execute lighthouse, unlighthouse-ci, and linkinator (Sections 4.7, 8.3, 15.2).
Evidence: Makes requests to well-known services including Google's PageSpeed Insights API (googleapis.com) and schema validators (Sections 4.6, 4.7.1).
[COMMAND_EXECUTION]: The skill frequently uses shell commands to perform its primary functions, including network probing and running audit tools.
Evidence: Extensive use of curl for fetching site data and metadata (Sections 8.1, 8.2, 8.4).
Evidence: Execution of npx and lhci commands for automated SEO auditing and CI gates (Sections 8.3, 15.2, 15.3).
[DYNAMIC_EXECUTION]: The skill utilizes dynamic execution environments via Node.js and shell scripts to parse data at runtime.
Evidence: Executes local Node.js scripts (e.g., scripts/parse-jsonld.mjs) to process fetched HTML content (Section 8.5).

web-seo