to-be-continuous-catalog

Warn

Audited by Gen Agent Trust Hub on May 16, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill requires the installation of a global NPM package gitlab-catalog-browser. This downloads code from the public NPM registry at runtime.
  • [COMMAND_EXECUTION]: The skill executes multiple shell commands to install the CLI, check versions, and interact with the GitLab catalog (e.g., npm install -g, gitlab-catalog-browser catalog list).
  • [REMOTE_CODE_EXECUTION]: The command gitlab-catalog-browser skills get to-be-continuous-catalog fetches workflows and reference material from a remote source and loads them into the agent's context. This constitutes dynamic loading of instructions from a remote repository managed by the CLI tool.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from the GitLab CI/CD component catalog (inputs, job definitions, schemas). While the skill includes instructions to treat this as data, malicious metadata in a public GitLab component could attempt to influence the agent's behavior.
  • Ingestion points: gitlab-catalog-browser catalog info, component inputs, component jobs, component workflows (SKILL.md).
  • Boundary markers: The skill explicitly instructs the agent to treat external content as DATA, not instructions.
  • Capability inventory: Global package installation (npm install -g), shell command execution, and remote instruction fetching.
  • Sanitization: No explicit sanitization or validation of the fetched component specifications is performed beyond the CLI's internal logic.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 16, 2026, 04:30 PM