The Agent Skills Directory

[DATA_EXFILTRATION]: The Python script scripts/openrouter_usage.py contains logic that attempts to read sensitive configuration files from hardcoded, highly specific paths on the user's local machine.
Evidence: The load_api_key function in scripts/openrouter_usage.py probes for .env files at ~/Desktop/Aldea/Prompt development/Aldea-Soul-Engine/.env and ~/Desktop/minoanmystery-astro/.env. It also checks ~/.config/env/global.env.
Impact: This behavior is irregular for a general-purpose utility and results in the reading of potentially sensitive files that may contain unrelated secrets, which are then loaded into the script's memory and could be exposed.
[PROMPT_INJECTION]: The skill processes data from an external API, creating a surface for indirect prompt injection.
Ingestion points: Usage and credit data are fetched from the OpenRouter API endpoints (/api/v1/activity and /api/v1/credits) in scripts/openrouter_usage.py.
Boundary markers: No boundary markers or delimiters are used to separate the external API data from the agent's instructions.
Capability inventory: The skill is limited to formatting and printing reports to the standard output; it does not possess capabilities for file modification or system command execution based on this data.
Sanitization: The script does not perform sanitization or validation of the API response content before formatting it into the usage report.

openrouter-usage