skills/datadrivenconstruction/ddc_skills_for_ai_agents_in_construction/specification-extractor/Gen Agent Trust Hub
specification-extractor
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from construction specification PDFs through the
pdfplumberlibrary and processes the extracted text to identify requirements and submittals. This surface could be exploited if an attacker embeds malicious instructions within a PDF to influence the agent's subsequent analysis or actions. - Ingestion points:
SpecificationExtractor.extract_from_pdfinSKILL.mdreads and extracts text from user-provided PDF files. - Boundary markers: The skill does not implement specific delimiters or 'ignore embedded instructions' warnings when processing the extracted text.
- Capability inventory: The skill possesses filesystem access (granted in
claw.json) and the ability to execute Python scripts, though no network operations were detected. - Sanitization: There is no evidence of sanitization or filtering of the extracted text before it is returned as part of the
SpecExtractionResultor used in generated reports.
Audit Metadata