process-db-report

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted external data to generate code and commands, creating an indirect prompt injection surface.
  • Ingestion points: The skill is designed to "Extract data from the pasted observability report" provided in the chat context.
  • Boundary markers: Absent. The instructions do not define clear delimiters or instruct the agent to ignore any embedded natural language instructions within the pasted reports.
  • Capability inventory: The skill has the authority to modify source files (specifically packages/db/src/schema.ts) and execute shell commands via bun run db:generate and bun run db:migrate.
  • Sanitization: Absent. There are no instructions for the agent to validate or sanitize the report content before interpolating it into the generated code snippets or action plans.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 04:55 AM