The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the platform's CLI tool to verify identity and workspace state (e.g., clay whoami) and to discover technical schemas for workflow actions (e.g., clay workflows actions list, clay workflows actions schema). It also involves reading and writing to local files referred to as 'answer sheets' to persist user-supplied identifiers and preferences.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could contain malicious instructions designed to manipulate the agent's behavior.
Ingestion points: The skill scrapes raw Google Search results using the ZenRows service, specifically targeting the bodyText field for parsing business information.
Boundary markers: There are no explicit delimiters or boundary instructions provided to help the agent distinguish between the scraped data and the skill's original instructions.
Capability inventory: The agent has access to shell command execution through the clay CLI, file system read/write access for preference management, and network capabilities via various data providers.
Sanitization: Although the skill mentions stripping known UI-related tokens from search results, it lacks robust sanitization to prevent the execution of instructions that might be embedded in the scraped text.
[EXTERNAL_DOWNLOADS]: Fetches data from external sources and well-known services, including Google Search (via ZenRows) and business location data from OpenMart. These services are used to build, deduplicate, and enrich the business lists requested by the user.

source-local-businesses