engineering-calc-report

Pass

Audited by Gen Agent Trust Hub on Nov 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user-provided calculations, scripts, and notes to generate formal, checkable reports.
  • Ingestion points: Gathered from the conversation, attached scripts/spreadsheets, or outputs from other skills (SKILL.md).
  • Boundary markers: The instructions do not specify explicit delimiters or warnings to ignore embedded instructions within user-provided data.
  • Capability inventory: The skill uses Python to write local files (calc_steps.json, calc_steps.md) and generates report artifacts.
  • Sanitization: The workflow mandates re-calculating values using the pint unit library via scripts/calc_template.py. This serves as a strong validation layer for numerical data, though it does not strictly sanitize instructional text injected into report prose.
  • [COMMAND_EXECUTION]: The Python template scripts/calc_template.py performs file write operations to generate report artifacts.
  • Evidence: The script writes structured data to calc_steps.json and markdown to calc_steps.md using standard file I/O operations. These operations are restricted to fixed filenames within the execution context.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 7, 2026, 06:02 AM