pipeline-operations

Warn

Audited by Gen Agent Trust Hub on Apr 28, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill provides instructions that access sensitive local file paths, specifically the user's SSH private key at ~/.ssh/id_rsa, which is passed to the pipelines start command for external repository integration.
  • [COMMAND_EXECUTION]: The skill makes use of shell command execution via the pipelines-cli and python3 to perform pipeline operations and data parsing. It also contains instructions for installing the CLI using sudo.
  • [EXTERNAL_DOWNLOADS]: The skill includes instructions to download a binary from https://cloud.acquia.com/pipelines-client/download, which is a domain associated with the vendor.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the processing of build outputs. Evidence Chain: 1. Ingestion points: Build logs and status metadata retrieved via pipelines logs and pipelines status. 2. Boundary markers: Absent in command examples. 3. Capability inventory: Shell command execution (pipelines CLI) and script execution (python3). 4. Sanitization: Not identified in the provided patterns for handling external build data.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 28, 2026, 01:51 PM