The Agent Skills Directory

[COMMAND_EXECUTION]: The skill employs shell commands for essential project workflows. Specifically, it performs git fetch and git pull --rebase to synchronize the local workspace with the remote origin repository in SKILL.md. It executes a local Python script scripts/audit_seo.py to automate the scanning of HTML and template files as seen in agents/auditor.md. It also suggests using package managers like npm to install framework-specific SEO utilities in agents/implementer.md.
[PROMPT_INJECTION]: The skill maintains an indirect prompt injection attack surface inherent to its auditing function. Ingestion points: The agent reads and processes untrusted HTML and template files (e.g., Vue, Astro, Svelte) from the user's local project directory via scripts/audit_seo.py and during manual review in agents/auditor.md. Boundary markers: There are no specialized delimiters or system instructions used to isolate the content of audited files from the agent's logic. Capability inventory: The skill has file-system write permissions (agents/implementer.md) and shell execution capabilities (SKILL.md), which are necessary for its primary function but represent an attack surface. Sanitization: The audit process parses HTML elements but does not explicitly filter for or sanitize potential prompt injection strings within the file content.

seo-ai-optimizer