The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a local Python script (scripts/score.py) to validate the business logic of generated offers. The execution is strictly limited to this specific script path through the YAML frontmatter's allowed-tools configuration, which effectively mitigates the risk of arbitrary shell command injection.- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process content from external URLs to identify potential product improvements for sales offers. Ingestion points: Content is retrieved from public buyer pages or product URLs as described in SKILL.md. Boundary markers: The instructions do not define specific delimiters to isolate external content from the agent's core instructions. Capability inventory: The skill's operations are confined to reading files, writing to the ./gtm/ project subdirectory, and running the provided scoring script; it has no network exfiltration or unrestricted file writing capabilities. Sanitization: There is no documented process for sanitizing or escaping the data retrieved from external web pages before it is interpolated into the generated offer text.

cold-offer