The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests untrusted data from project context files and web search results to inform its outreach generation. While this is inherent to the skill's purpose, the external content is interpolated into agent prompts without explicit sanitization or boundary markers.
Ingestion points: The skill reads from projects/<project>/gtm-context.md, icp-personas.md, competitive-intel.md, and positioning.md, and processes results from the WebSearch tool.
Boundary markers: There are no specific delimiters or "ignore instructions" warnings used when providing this ingested data to the sub-agents in Step 2.
Capability inventory: The skill uses Read, Write, WebSearch, and Task tools across its scripts.
Sanitization: No escaping or validation is applied to the content retrieved from external files or web searches before it is used for sequence generation.

gtm-outreach