create-pipeline-v1

Pass

Audited by Gen Agent Trust Hub on Jun 15, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it transforms untrusted user requirements into executable CI/CD configurations and possesses the capability to write these configurations to a remote platform.
  • Ingestion points: User-provided requirements for pipeline logic, build steps, and environment configurations enter the agent's context via the chat interface.
  • Boundary markers: Absent. The skill's instructions do not define delimiters or provide specific 'ignore embedded instructions' warnings for the generated YAML output to prevent the agent from obeying instructions hidden within user input.
  • Capability inventory: The skill utilizes the harness_create tool (facilitated by the harness-mcp-v2 server) to push generated configurations to the Harness platform, as described in the 'Creating via MCP' section of SKILL.md.
  • Sanitization: Absent. There are no defined procedures for validating, escaping, or sanitizing the content of user-provided requirements before they are interpolated into the run or action steps of the generated pipeline.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 15, 2026, 10:16 AM
Security Audit — agent-trust-hub — create-pipeline-v1