The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes curl to perform network requests to the Brand.dev API and python3 -c to parse the resulting JSON data. This is standard functionality for an API integration skill and does not involve dynamic or untrusted code execution.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from external web sources (brand mentions, titles, and snippets) via the Brand.dev API endpoints. This creates a surface for indirect prompt injection where malicious instructions embedded in web content could attempt to influence the agent's reporting or behavior.
Ingestion points: Brand mentions returned from the /v1/brand/search and /v1/monitors/{monitor_id}/mentions endpoints in SKILL.md.
Boundary markers: No explicit delimiters or instructions to ignore embedded content are used when the external data is processed.
Capability inventory: The skill has capabilities for shell command execution (curl) and network requests to the API domain.
Sanitization: The provided scripts do not perform sanitization or filtering of the mention text before presenting it to the agent for analysis.

brand-monitor