import-duty-check-free
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted tax document text and JSON payloads as primary input to the execution engine. While the current implementation uses deterministic arithmetic logic, the ingestion of external data without explicit boundary markers in the
scripts/run.mjswrapper represents an attack surface where maliciously crafted document text could attempt to influence the agent's interpretation of results. - Ingestion points:
scripts/run.mjsvia--inputparameter (file ingestion) andENGINE.run(payload)inscripts/engine/import-duty-check.js. - Boundary markers: Absent. The input text is processed directly as a string or parsed JSON without wrapping delimiters or 'ignore embedded instructions' prompts.
- Capability inventory: The skill uses
ReadandBashtools.scripts/run.mjsperforms local file reads (fs.readFileSync) and standard output. - Sanitization: The engine performs basic normalization (
normNumber) and structural validation (parseTable), but does not sanitize the input text for embedded natural language instructions before it is presented to the agent context via output.
Audit Metadata