The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions direct the user to clone a repository from an unverified external source: https://github.com/vishalGitthub/cli-llm-mesh.git.
[REMOTE_CODE_EXECUTION]: The setup process involves executing an unverified bootstrap script (python bootstrap.py) immediately after downloading the repository content, which is a high-risk operation as the script's contents are not inspected.
[COMMAND_EXECUTION]: The skill relies on several shell commands for installation and operation, including pip install for unknown dependencies and python firepulse.py for credential management and interactive sessions.
[PROMPT_INJECTION]: The skill provides a surface for indirect prompt injection because it orchestrates data from multiple external LLM providers and user-supplied queries without documented sanitization.
Ingestion points: Data enters the system via command-line arguments (the --query flag) and via the responses fetched from third-party AI APIs (xAI, OpenRouter, Mistral, DeepSeek).
Boundary markers: There are no explicit delimiters or instructions provided to the agent to disregard malicious instructions embedded within the data retrieved from external APIs.
Capability inventory: The skill possesses the ability to read and write to the local file system (specifically the ~/.firepulse directory) and perform network requests to various external services.
Sanitization: There is no evidence of input validation, output escaping, or filtering of the content processed by the orchestrator.

firepulse-multi-model-ai-orchestrator