launch-economics
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external JSON files which could potentially contain malicious instructions that could influence the agent when it reads the report output.\n
- Ingestion points: Data is read from files specified via CLI arguments in
scripts/launch_economics.py(via--input) andscripts/viability_verdict.py(via--economicsand--evidence).\n - Boundary markers: The instructions in
SKILL.mdand the output template inassets/viability-memo-template.mdprovide structural boundaries for the data, but no explicit delimiters or escape sequences are used to isolate potentially malicious string content provided in the input JSON.\n - Capability inventory: The skill scripts are limited to local file reading (JSON parsing), mathematical computation, and text formatting. They do not possess capabilities for network communication, file system modification, or shell command execution.\n
- Sanitization: The scripts implement schema validation for data types and required fields, but they do not sanitize string content (such as business names or unit labels) for potential prompt injection patterns before including them in the report output.
Audit Metadata