The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the hermes-agent package from the NousResearch GitHub repository. This is a well-known organization in the artificial intelligence research community.
[COMMAND_EXECUTION]: The skill executes a wrapper script (run_x_search.py) using a specific Python interpreter located in a local virtual environment created by the uv tool. This is a standard method for running CLI-based tools.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes and summarizes content fetched from X (Twitter). Maliciously crafted posts could potentially attempt to influence the agent's behavior.
Ingestion points: Output of run_x_search.py containing search results from Grok.
Boundary markers: None identified in the script's output wrapper.
Capability inventory: The skill uses standard shell execution for the search operation.
Sanitization: No explicit sanitization of the search results is performed before they are returned to the agent.
[DATA_EXFILTRATION]: While the skill instructions discuss the location and management of xAI credentials (auth.json), it does so to guide the user through a legitimate authentication process for the Hermes Agent tool and does not appear to access or transmit these credentials to unauthorized locations.

x-search