ship-workflow-canary
Fail
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [REMOTE_CODE_EXECUTION]: Automated security scans identified a high-risk pattern where content fetched from a remote URL is piped to a shell or utility. In the provided scripts, curl output is piped to shasum. While intended for integrity checking, piping untrusted remote data into command-line tools without strict validation is a security risk.\n- [COMMAND_EXECUTION]: The skill's health check and baseline scripts interpolate the ${HOST} and ${endpoint} variables directly into bash commands using subshells. This creates a vulnerability to shell command injection if an attacker can influence these variables through indirect prompt injection or other means. Evidence: 1. Ingestion points: ${HOST} and ${endpoint} variables. 2. Boundary markers: None. 3. Capability inventory: curl, shasum, subshell execution. 4. Sanitization: None.\n- [EXTERNAL_DOWNLOADS]: The skill initiates network requests to external endpoints using curl to perform health checks. This capability could be leveraged to interact with unintended or malicious remote servers if the host variable is manipulated.
Recommendations
- HIGH: Downloads and executes remote code from: https://${HOST}${endpoint} - DO NOT USE without thorough review
Audit Metadata