The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes content from third-party websites (e.g., brand sites, competitor URLs, and Reddit) to extract terminology, positioning, and content gaps. This creates a surface for indirect prompt injection where malicious instructions embedded in those websites could influence the agent's behavior.
Ingestion points: External website content and search results are ingested during Phase 1 (Brand Intelligence), Phase 2 (Keyword Discovery), and Phase 4 (Competitor Analysis) as defined in SKILL.md.
Boundary markers: The instructions do not define explicit boundary markers or 'ignore instructions' delimiters for the content extracted from external URLs.
Capability inventory: The agent possesses file-writing capabilities (keywords.csv), command execution (running local .mjs scripts), and network access (via WebFetch).
Sanitization: There is no explicit sanitization of the text extracted from external websites before it is processed by the agent.
[COMMAND_EXECUTION]: The skill instructions in SKILL.md (Phase 2) direct the agent to execute internal Node.js scripts (keyword-explorer.mjs and serp-analyzer.mjs) via the shell to supplement research data with real SERP metrics. While these scripts are provided with the skill, this pattern represents an intended use of local command execution capabilities.
[SAFE]: The skill performs network operations to fetch data from well-known and legitimate services, including Google and SerpAPI. These operations are conducted to fulfill the primary purpose of keyword discovery and do not involve access to sensitive local files or credentials.

research-keywords