The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute curl commands to interact with the orq.ai API for fetching and updating prompt versions.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted prompt text from external sources.
Ingestion points: Prompts are fetched from the orq.ai API via search_entities and GET requests, or provided directly by the user.
Boundary markers: The instructions specify preserving template variables like {{variable_name}}, but do not employ strict delimiters or clear "ignore embedded instructions" warnings for the content under analysis.
Capability inventory: The agent has access to Bash (allowing network exfiltration via curl), Write, Edit, and various orq platform tools.
Sanitization: No explicit sanitization or structural escaping is performed on the input prompt before it is analyzed by the LLM, though the skill mandates a human-in-the-loop review of the diff before any changes are applied.

optimize-prompt