manage-agent
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches the amctl installation script from the vendor's official GitHub repository (github.com/wso2/agent-manager).
- [REMOTE_CODE_EXECUTION]: Executes the remote installation script using a piped shell command (curl | sh). This is the standard, vendor-provided installation method for the CLI tool.
- [COMMAND_EXECUTION]: Invokes the amctl CLI and local wrapper scripts to manage agent lifecycles, including creation, deployment, and retrieval of logs and metrics.
- [PERSISTENCE]: The installation script adds the amctl binary to the system PATH by updating shell configuration files (e.g., .bashrc, .zshrc), which is a standard procedure for CLI tool availability across sessions.
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for reading external agent logs and traces, creating a surface for potential untrusted data ingestion. * Ingestion points: Processes data from 'amctl agent logs', 'amctl agent build logs', and 'amctl agent traces'. * Boundary markers: No explicit delimiters are used to separate log content from instructions. * Capability inventory: The skill maintains full lifecycle management capabilities for agents via the amctl CLI. * Sanitization: The skill does not explicitly specify sanitization of log or trace output before processing.
Audit Metadata