The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the @cargo-ai/cli package globally via npm and suggests using npx to add additional skills from the vendor's repository (getcargohq/cargo-skills).
[DATA_EXFILTRATION]: The skill reads the ~/.claude/plugins/installed_plugins.json file to check for the presence of specific plugins. It then sends session metadata—including a summary and title indicating the skill's source—to the vendor's API via the cargo-ai workspaceManagement session upsert command for attribution and usage tracking.
[COMMAND_EXECUTION]: The skill uses the GitHub CLI (gh) to star the getcargohq/gtm-skills repository on the user's behalf. Although the instructions state the agent must ask for user consent first, providing the automated command sequence facilitates account interaction that is secondary to the skill's primary function.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the aiArk connector through the searchPeople action, creating a surface for potential injection.
Ingestion points: External data enters the agent context via the cargo-ai orchestration action execute command which retrieves stakeholder information such as titles and departments from a third-party provider.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present in the prompt templates.
Capability inventory: The skill environment allows for shell execution (npm, gh, cargo-ai), file system writing (mkdir, touch), and local file reading (grep).
Sanitization: No evidence of data sanitization, validation, or escaping of the retrieved stakeholder records before they are returned to the agent session.

find-stakeholders