ironbee-backend-devtools-cli
Pass
Audited by Gen Agent Trust Hub on May 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on the execution of the
ironbee-backend-devtools-clibinary, which can interact with local and remote resources, such as reading log files from path-based sources and managing Docker or Kubernetes logs. - [EXTERNAL_DOWNLOADS]: Documentation instructs the installation of the
@ironbee-ai/devtoolsnpm package, which is a resource managed by the skill's author. - [REMOTE_CODE_EXECUTION]: The
executeandscenariotools support running JavaScript within a virtual machine (VM) for batch operations, enabling dynamic code execution at runtime. - [DATA_EXFILTRATION]: The
requestsuite (HTTP, gRPC, GraphQL, WebSocket) allows the agent to send data to arbitrary external URLs. This functionality is intended for testing but provides a potential channel for data exfiltration. - [PROMPT_INJECTION]: The skill processes untrusted data from external logs, database records, and network responses. While this presents an indirect prompt injection surface, the tool implements mitigations including automatic redaction of sensitive keys like 'password', 'token', and 'api_key'.
Audit Metadata