The Agent Skills Directory

[SAFE]: The skill is a set of instructional templates and recipes designed to guide an agent in using the Cargo platform's GTM tools. All external service interactions (e.g., SalesNavigator, Waterfall, PeopleDataLabs) are established GTM providers and are used as intended for the skill's primary purpose.
[COMMAND_EXECUTION]: The skill makes extensive use of the @cargo-ai/cli tool to perform batch data processing, enrichment, and storage operations. These commands are well-documented and involve standard shell utilities like jq and curl for managing data flow between platform actions.
[DATA_EXPOSURE]: The skill processes business contact information (emails, phone numbers, LinkedIn URLs) and firmographic data. This behavior is consistent with the stated purpose of a GTM prospecting tool and is performed through authenticated vendor connectors.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it interpolates data enriched from external sources (such as LinkedIn profiles) into LLM prompts for lead scoring and personalization.
Ingestion points: Enriched company and prospect data fetched from third-party APIs (Waterfall, PeopleDataLabs, etc.) are used in prompts within guides/writing-outreach.md and recipes/icp-discovery.md.
Boundary markers: Prompt templates use simple curly-brace interpolation (e.g., {firmographics}) without explicit boundary delimiters or instructions to ignore embedded commands.
Capability inventory: The agent has capabilities to execute platform-specific commands via cargo-ai for data storage, synchronization with CRMs (HubSpot, Salesforce), and further enrichment calls.
Sanitization: There is no evidence of explicit sanitization or validation of the external content before it is processed by LLM actions.

cargo-gtm