The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from local project files to provide contextualized marketing advice, which creates an attack surface for indirect prompt injection if those files are controlled by a malicious actor.
Ingestion points: Instructions in SKILL.md prompt the agent to read .agents/product-marketing.md, .claude/product-marketing.md, or product-marketing-context.md before proceeding.
Boundary markers: The skill lacks explicit boundary markers or instructions to disregard potential system-level commands embedded within the context files.
Capability inventory: The skill is primarily instructional and generates marketing plans, copy, and platform recommendations. It does not have built-in capabilities for file system modification or network requests.
Sanitization: There is no evidence of sanitization or validation of the content read from the external context files.
[EXTERNAL_DOWNLOADS]: The skill provides references to an external tools registry and integration guides hosted on a third-party GitHub repository.
Evidence: SKILL.md contains links to https://github.com/coreyhaines31/marketingskills/blob/main/tools/REGISTRY.md and several related sub-pages for implementation details.
Note: While these are documented as resources for the user, an agent following these links would be accessing content from an unverified personal repository.

sms