The Agent Skills Directory

[PROMPT_INJECTION]: The instructions direct the agent to prioritize reading external context files (".agents/product-marketing.md", ".claude/product-marketing.md", or "product-marketing-context.md") before interacting with the user. If these files contain malicious directives, they could be used to override the skill's primary instructions or bypass safety guidelines.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection due to the processing of untrusted external data without sufficient protections.
Ingestion points: The skill reads local context files and instructs the agent to analyze call transcripts and buyer interviews during win-loss analysis (SKILL.md, references/win-loss-analysis.md).
Boundary markers: The skill lacks instructions to use delimiters or "ignore embedded instructions" warnings when processing these external inputs.
Capability inventory: The agent possesses capabilities to generate extensive text and files, including sales decks, playbooks, and demo scripts.
Sanitization: There is no mention of sanitizing or validating the content of the transcripts or context files before they are used to influence the agent's output.
[EXTERNAL_DOWNLOADS]: The skill references a tool registry and integration guides hosted on a public GitHub repository at "github.com/coreyhaines31/marketingskills". Although GitHub is a well-known service, these resources originate from an external account not associated with the primary vendor.

sales-enablement