The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes the reportlab library for PDF generation, which is a standard and well-regarded Python package.\n- [COMMAND_EXECUTION]: The skill executes a Python script to process account data and generate reports; these operations are within the expected scope of the tool's functionality.\n- [DATA_EXFILTRATION]: Data is retrieved from api.recoupable.com, which is the official API domain for the 'recoupable' vendor. Reports are saved to a local directory as requested by the user.\n- [CREDENTIALS_UNSAFE]: Authentication is handled via a standard environment variable (RECOUP_ACCESS_TOKEN), ensuring that no sensitive credentials are hardcoded in the script.\n- [PROMPT_INJECTION]: The skill ingests data from the Recoup API (such as chat titles) to populate reports. While this creates an indirect prompt injection surface, it is a necessary part of the tool's reporting function, and the script includes sanitization for file paths and Markdown output.\n
Ingestion points: Account activity, artist names, and chat titles are fetched from the Recoup API in scripts/generate_account_status.py.\n
Boundary markers: The generated Markdown and PDF files structure the data into clear fields, although they do not use specific 'ignore instructions' delimiters for the values.\n
Capability inventory: The tool performs file write operations to the local disk.\n
Sanitization: A slug() function is used for path safety, and chat titles are truncated and escaped for Markdown table safety.

recoup-internal-account-health-report