The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The file scripts/desform_jimureport.py contains a hardcoded signature secret (dd05f1c54d63749eda95f9fa6d49v442a) used for signing API requests.\n- [COMMAND_EXECUTION]: User-provided strings for form names, codes, and titles are interpolated into SQL statements and shell commands without escaping. The gen_menu_sql function in scripts/desform_utils.py is particularly vulnerable, as it uses simple string formatting for INSERT statements, potentially allowing SQL injection if the scripts are executed in a database environment.\n- [INDIRECT_PROMPT_INJECTION]: The skill facilitates processing untrusted data to generate complex configurations and code, presenting an injection risk.\n
Ingestion points: Form requirements and data records processed via desform_data_utils.py.\n
Boundary markers: None; there are no delimiters or instructions to ignore embedded commands when processing user data.\n
Capability inventory: Includes network requests via urllib.request and shell command generation for database tasks.\n
Sanitization: Relies on JSON serialization for payloads but lacks escaping for generated SQL scripts.\n- [DYNAMIC_CONTEXT_INJECTION]: The SKILL.md file uses the !command`` syntax to execute scripts/gen_widget_options_index.py at load time, allowing pre-execution shell command invocation.\n- [DATA_EXFILTRATION]: In scripts/desform_utils.py, the SSL context is configured to skip certificate verification (ssl.CERT_NONE), which could lead to sensitive data exposure via man-in-the-middle (MITM) attacks.

jeecg-desform