generate-validation-notebook

Pass

Audited by Gen Agent Trust Hub on Apr 8, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands using gh (GitHub CLI) and git to retrieve repository metadata and file contents. It also runs local Python scripts and invokes the open command to launch the generated notebook URL in the user's browser.
  • [DATA_EXFILTRATION]: The skill transmits dbt model definitions and SQL queries to the Monte Carlo platform by encoding them into a URL fragment. This is a functional requirement for the notebook import feature, and the destination domain (getmontecarlo.com) aligns with the author's identity ('monte-carlo-data').
  • [PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes untrusted data from GitHub PRs (diffs, comments, and file contents).
  • Ingestion points: Data enters the context via gh pr view, gh pr diff, and gh api calls in SKILL.md (Phase 1).
  • Boundary markers: The instructions do not define explicit delimiters to isolate untrusted content from the agent's logic.
  • Capability inventory: The skill can execute shell commands (gh, git, python3) and write to the local filesystem (/tmp/validation_notebook_working/).
  • Sanitization: While helper scripts use yaml.safe_load(), there is no explicit validation or sanitization of the SQL code extracted from PR diffs before it is incorporated into the generated notebook output.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 8, 2026, 06:52 PM