cisco-uc-engineer
Pass
Audited by Gen Agent Trust Hub on Apr 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes various CLI tools (e.g., cisco-axl, cisco-dime) to perform system administration tasks, including the capability to run raw SQL queries against the Cisco Unified Communications Manager (CUCM) database using the
cisco-axltool. - [CREDENTIALS_UNSAFE]: The instructions direct the agent to access and interact with configuration files located at paths such as
~/.cisco-support/config.json,~/.audiocodes-cli/config.json, and~/.genesys-cli/config.json, which contain sensitive credentials including API keys, OAuth tokens, and system passwords. - [EXTERNAL_DOWNLOADS]: The skill documentation recommends the installation of global npm packages from the public registry (e.g.,
cisco-dime) for core functionality, which introduces external dependencies into the agent's execution environment. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it ingests untrusted data from system syslogs, SIP traces, and Call Detail Records (CDRs) which are then processed by the agent. Ingestion points:
cisco-dime(syslogs, SIP traces) andcisco-cdr(call search/trace). Boundary markers: None specified in the instructions for log processing. Capability inventory: Includes SQL execution viacisco-axl, configuration changes, and file system downloads viacisco-dime. Sanitization: No explicit sanitization or validation of the ingested log content is provided.
Audit Metadata