The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Interacts with the vendor's official domains (api.orq.ai, docs.orq.ai, and my.orq.ai) to fetch prompt data and documentation. These operations are essential for the skill's functionality and target trusted infrastructure.
[COMMAND_EXECUTION]: Provides instructions for using curl to interact with the orq.ai HTTP API as a fallback. These commands correctly utilize environment variables ($ORQ_API_KEY) for authentication rather than hardcoding secrets.
[PROMPT_INJECTION]: The skill processes untrusted system prompts during the analysis and optimization phase, presenting a surface for indirect prompt injection. However, it incorporates mitigation strategies such as mandatory diff reviews, explicit instructions to preserve original intent, and strict preservation of template variables ({{variable_name}}).
Ingestion points: Reads system messages from the orq.ai API or directly from user input in SKILL.md workflows.
Boundary markers: Not explicitly defined for the prompt content, but instructions emphasize literal preservation of variables.
Capability inventory: Uses Bash, Write, Edit, and WebFetch tools to manage local files and API communication.
Sanitization: Relies on human-in-the-loop review and diff validation before any changes are applied.

optimize-prompt