The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the ingestion of external context files.
Ingestion points: The skill directs the agent to read content from .agents/product-marketing.md, .claude/product-marketing.md, or product-marketing-context.md before proceeding with community strategy tasks.
Boundary markers: There are no instructions to wrap these external inputs in delimiters (like XML tags) or to explicitly ignore any instructions or overrides that might be contained within those files.
Capability inventory: The agent uses the data from these files to generate strategic documents, including community plans, channel architectures, and ambassador program briefs.
Sanitization: The skill lacks any mechanism to validate or sanitize the content of the context files, which could lead to the agent following malicious instructions embedded in those files.

community-marketing