The Agent Skills Directory

[PROMPT_INJECTION]: No evidence of instructions attempting to bypass safety filters, override agent behavior, or extract system prompts was found.
[DATA_EXFILTRATION]: The skill requires access to a company URL and client context to perform SEO analysis. There are no patterns indicating that this sensitive data is being sent to unauthorized external domains.
[COMMAND_EXECUTION]: The playbook describes a high-level workflow and does not contain any direct shell command executions, privilege escalation attempts, or persistence mechanisms.
[REMOTE_CODE_EXECUTION]: No remote scripts or unverified dependencies are downloaded or executed at runtime. The installation method uses a standard package manager command.
[INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection as it ingests untrusted data from external company websites and client-provided context files.
Ingestion points: company-url and client-context parameters in SKILL.md.
Boundary markers: None identified in the prompt templates.
Capability inventory: Orchestrates sub-skills such as seo-content-audit and content-asset-creator to process and generate content.
Sanitization: No explicit sanitization or filtering of the ingested content is described in this playbook.

seo-content-engine