bim-to-schedule-4d

Pass

Audited by Gen Agent Trust Hub on Mar 5, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements standard construction data logic using standard Python libraries such as pandas and datetime. All logic is self-contained and operates on user-provided data.
  • [DATA_EXFILTRATION]: The skill includes a method to export results to the local filesystem via export_links in SKILL.md. This functionality is consistent with the filesystem permission declared in claw.json and is intended for generating Excel reports.
  • [PROMPT_INJECTION]: The skill ingests external datasets through the import_schedule and import_elements methods in SKILL.md. While this constitutes a data ingestion surface, the risk of indirect prompt injection is mitigated by the instruction to 'Validate inputs before processing' and the absence of high-risk capabilities like network access or shell execution. Ingestion points: import_schedule and import_elements in SKILL.md. Boundary markers: Absent in prompt instructions. Capability inventory: Local filesystem write via export_links in SKILL.md. Sanitization: Instructions explicitly mandate input validation before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 5, 2026, 04:28 AM