ci-gen

Warn

Audited by Gen Agent Trust Hub on Mar 4, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to run npx ai-ci, which executes a command-line tool from the public npm registry. This grants the tool the ability to execute arbitrary code on the host system.
  • [EXTERNAL_DOWNLOADS]: The ai-ci package is downloaded at runtime from the npm registry. The package owner (LXGIC Studios) is not listed as a trusted vendor for the skill author (sundial-org), making this an external dependency from an untrusted source.
  • [REMOTE_CODE_EXECUTION]: By using npx to execute unversioned code from a remote registry, the skill facilitates remote code execution where the behavior of the tool could be altered at the source without oversight.
  • [PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests project data to generate output.
  • Ingestion points: Local project files including package.json, configuration files, and overall project structure.
  • Boundary markers: Absent. The instructions do not define clear delimiters or safety prompts to prevent the agent from following instructions embedded in the scanned project files.
  • Capability inventory: The skill can read project files, write new workflow files, and execute shell commands via npx.
  • Sanitization: No sanitization or validation of project-specific content is mentioned before it is processed by the AI logic.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 4, 2026, 02:22 AM