company-knowledge

Pass

Audited by Gen Agent Trust Hub on Jul 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local shell script to determine the root directory for the company knowledge corpus.
  • Evidence: SKILL.md contains an instruction to execute sh scripts/resolve-root.sh to resolve the environment root.
  • [PROMPT_INJECTION]: The skill reads and processes external Markdown files from a locally configured directory, which creates a surface for indirect prompt injection.
  • Ingestion points: Markdown files located within the directory resolved by the resolve-root.sh script or the COMPANY_KNOWLEDGE_ROOT environment variable.
  • Boundary markers: None explicitly defined; there are no instructions to disregard commands or prompts embedded within the external knowledge files.
  • Capability inventory: The agent has shell access via the skill's own script and standard file-system operations.
  • Sanitization: No sanitization or content validation is performed on the Markdown data ingested from the knowledge base.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 17, 2026, 08:57 AM
Security Audit — agent-trust-hub — company-knowledge