The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill possesses a significant data ingestion surface as it pulls information from external sources including Ahrefs, web search results via agent-browser, and client-provided materials. The methodology does not explicitly define boundary markers or sanitization procedures for this untrusted content, creating a surface for potential indirect prompt injection attacks.
[DATA_EXFILTRATION]: The skill is designed to access and aggregate highly sensitive business data, including financial metrics from Stripe (MRR, ARR, LTV), customer engagement data from Customer.io, and internal repository structures from GitHub. While these operations are functional for strategy generation and intended to be stored in the user's local ~/marketing-plans/ directory, the concentration of such data in a single research file creates a high-value target for exfiltration if the agent context is compromised.
[EXTERNAL_DOWNLOADS]: The ops-stack-mapping.md documentation recommends the manual installation of external command-line tools such as agent-browser and defuddle using npm. These tools are intended to facilitate data extraction and browser automation, representing external dependencies that the user is encouraged to integrate into their environment.

marketing-plan