write-context-rules
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill represents a surface for indirect prompt injection because it ingests untrusted data from a local repository and incorporates it into a central
RULES.mdfile, which is loaded as system-level context for every subsequent agent interaction.\n - Ingestion points: The skill reads project metadata, schemas, and free-text documentation from
nao_config.yaml,databases/,repos/, anddocs/directories (Source: SKILL.md).\n - Boundary markers: The generated output in
RULES.mdlacks explicit boundary markers or 'ignore' directives to prevent the agent from accidentally obeying instructions that might be embedded within the project's documentation or table descriptions.\n - Capability inventory: The skill is capable of performing file system writes to
RULES.mdand executing web searches to gather company and business model information (Source: SKILL.md).\n - Sanitization: No automated sanitization, filtering, or validation is performed on the data extracted from the repository before it is integrated into the core instruction set.
Audit Metadata