fayanruju
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill installation guide (INSTALL.md) references a vendor-controlled repository (15810879921-coder/oneos-pm-skills) for adding the skill using npx. This represents normal vendor deployment functionality according to the provided author context.\n- [COMMAND_EXECUTION]: The SKILL.md file defines a protocol for the agent to use a local read-only probe script (~/oneos-prod/scripts/prod-ro-probe.sh) to query production database facts. This capability is intended for the OneOS environment and includes safety guidelines to prevent credential exposure or write operations.\n- [DATA_EXPOSURE]: The testing script scripts/smoke-system-qa.py contains a hardcoded absolute path (/Users/sylvawong/oneos-v2/...) for locating local resources. This is assessed as a developer environment reference rather than a security risk.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process extensive external knowledge base content.\n
- Ingestion points: Markdown, JSON, and TSV files within the knowledge base directory.\n
- Boundary markers: The skill employs confidence levels and mandatory citation protocols.\n
- Capability inventory: Agent capabilities are focused on file reading and grep-based searching.\n
- Sanitization: Content is utilized for business rule adjudication without specific mention of interpolation sanitization.
Audit Metadata