The Agent Skills Directory

[SAFE]: No malicious behaviors or security vulnerabilities were identified. The implementation follows standard protocols for identity and access management.\n- [DATA_EXFILTRATION]: Network communication in the audit script is limited to the user-provided Okta organization domain. No evidence of unauthorized data transfer to third-party services was found.\n- [CREDENTIALS_UNSAFE]: Sensitive information such as API tokens is managed via environment variables and placeholders, adhering to secure credential handling practices.\n- [INDIRECT_PROMPT_INJECTION]:\n
Ingestion points: The SCIM server in scripts/process.py processes JSON input from HTTP requests for user and group management.\n
Boundary markers: Uses standard REST API structures and JSON schemas to define data boundaries.\n
Capability inventory: The skill facilitates database record creation and management; it does not utilize high-risk sinks like subprocess calls or dynamic execution.\n
Sanitization: Input is processed through SQLAlchemy ORM and structured parsing, mitigating common injection risks.

implementing-scim-provisioning-with-okta