update-key
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
curltool to interact with the LiteLLM proxy API. It performs GET requests to list keys and POST requests to update key configurations. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user-provided input (such as
max_budget,models, andduration) and interpolate these values into a JSON payload for an API request. This creates a potential surface for indirect prompt injection if the inputs are not validated by the underlying proxy service. - Ingestion points: User input requested for key modification parameters in
SKILL.md. - Boundary markers: Not explicitly defined in the prompt instructions for the JSON body construction.
- Capability inventory: Network operations via
curlto user-defined proxy endpoints. - Sanitization: None described in the skill instructions; the agent is expected to pass the values directly into the shell command.
Audit Metadata