The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The file scripts/system_utils.py contains a hardcoded cryptographic secret (_SIGN_SECRET) used for generating signatures for specific API requests. Hardcoding secrets in source code is an unsafe practice that exposes credentials to anyone with access to the skill's code.
[DATA_EXFILTRATION]: The skill handles sensitive X-Access-Token JWTs and transmits them to an API base URL provided via command-line arguments. This behavior can be exploited to harvest credentials if an attacker tricks a user into providing a malicious endpoint.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from external sources (JSON configuration files and command-line arguments) and uses it to perform network operations.
Ingestion points: Data enters the system via the --config JSON file and various CLI parameters in scripts/system_creator.py.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to prevent the execution of malicious instructions that might be embedded in the processed data.
Capability inventory: The skill has the ability to perform HTTP GET and POST requests to arbitrary URLs using the urllib.request library in scripts/system_utils.py.
Sanitization: The skill performs minimal sanitization (standard URL quoting) but does not validate the integrity or safety of the content before it is used in API calls.

jeecg-system