cs-onboard

Pass

Audited by Gen Agent Trust Hub on Jul 15, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface (Category 8) by collecting and persisting user-supplied data without sanitization.
  • Ingestion points: User-provided interview responses are processed and stored by instructions in SKILL.md via the /cs:setup and /cs:update commands.
  • Boundary markers: Although the output is structured using a template (assets/company-context-template.md), there are no explicit boundary markers or instructions to the agent to ignore potentially malicious content within the captured data fields.
  • Capability inventory: The skill performs file-write operations to ~/.claude/company-context.md (SKILL.md), which is a persistent file intended to be read by other skills.
  • Sanitization: No validation, escaping, or filtering is performed on the input before it is written to the persistent context file.
  • [COMMAND_EXECUTION]: The skill performs persistent file-system writes to the ~/.claude/ directory to store sensitive company data, including revenue, runway, and internal organizational challenges. This persistent data collection creates a high-value target for other potentially malicious skills on the same system.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 15, 2026, 07:04 PM
Security Audit — agent-trust-hub — cs-onboard