skills/filipas123/kimi-atlas/sun-path/Gen Agent Trust Hub

sun-path

Pass

Audited by Gen Agent Trust Hub on Nov 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill requests shell:exec permissions to execute a suite of local Python scripts, including sun_calc.py, plot_sunpath.py, shadow_calc.py, annual_sun_hours.py, terrain_shadow.py, and comfort_calc.py. These scripts are responsible for all geometry, solar, and environmental calculations.
  • [INDIRECT_PROMPT_INJECTION]: The skill exposes an attack surface for indirect prompt injection by processing external data and interpolating it into shell commands. Crucially, the instructions in SKILL.md direct the agent to "run the corresponding script directly" and "do not ask for confirmation," which bypasses the human-in-the-loop security control that would normally allow a user to inspect the command before execution.
  • Ingestion points: User-supplied parameters for geographic coordinates (latitude/longitude), building dimensions, and local file paths for GeoTIFF Digital Elevation Models (DEMs) as described in the SKILL.md usage examples.
  • Boundary markers: Absent; the skill does not provide instructions to use delimiters or ignore embedded instructions when processing user or file-based data.
  • Capability inventory: The skill relies entirely on shell:exec to perform its core analytical functions via subprocess calls to Python.
  • Sanitization: Absent; no explicit validation, escaping, or sanitization logic is provided to ensure that inputs do not contain shell metacharacters or command injection payloads.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 22, 2026, 01:12 AM