skills/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/document-classification-nlp/Gen Agent Trust Hub
document-classification-nlp
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted text from construction documents (RFIs, change orders, etc.), creating a surface for indirect instructions to be passed to the agent.
- Ingestion points: Document content passed to
process_document_batch,ConstructionDocumentClassifier.classify, andConstructionInfoExtractor.extract_allwithinSKILL.md. - Boundary markers: The implementation does not include specific boundary markers or instructions to ignore embedded commands within the text being analyzed.
- Capability inventory: The skill possesses
filesystempermissions and includes code for writing data to the local file system usingpandas.to_excelinSKILL.md. - Sanitization: There is no evidence of text sanitization or validation to filter out potential prompt injection strings within the document content before it is processed.
- [COMMAND_EXECUTION]: The skill implementation includes logic for writing files to the local system.
- Evidence: The
process_document_batchfunction inSKILL.mdusesdf.to_excel(output_path, index=False)to save processed results. - Context: This behavior is consistent with the skill's stated purpose of document management and reporting.
Audit Metadata