add-org
Pass
Audited by Gen Agent Trust Hub on Apr 6, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
Bashtool withcurlto perform a POST request to a user-specified LiteLLM proxy endpoint inSKILL.md. - [DATA_EXFILTRATION]: The skill requests a proxy admin key (
LITELLM_API_KEY) and base URL from the user to authorize API requests. This information is used to interact with the user's LiteLLM proxy as part of the skill's management functionality. - [PROMPT_INJECTION]: The skill processes user-supplied inputs such as organization name, models, and budget to construct a JSON payload for the
curlcommand. - Ingestion points: User-provided organization alias, model list, and budget limits.
- Boundary markers: None present; the skill uses placeholders within a JSON string inside a bash command.
- Capability inventory: Execution of shell commands via the
Bashtool inSKILL.md. - Sanitization: Relies on the agent to correctly format and insert values into the provided JSON template without explicit validation steps.
Audit Metadata