bucle-agentico
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's 'Context Mapping' phase involves reading existing files, codebase patterns, and database structures. This ingestion of external data creates an attack surface for indirect prompt injection if the analyzed files contain instructions designed to manipulate the agent's behavior.
- Ingestion points: Codebase (Grep/Glob/Read), Supabase (list_tables, execute_sql).
- Boundary markers: Not explicitly defined in the prompt instructions for external data ingestion.
- Capability inventory: File writing (TodoWrite), database modifications (Supabase apply_migration/execute_sql), error monitoring (Next.js), and UI interaction (Playwright).
- Sanitization: The skill recommends Zod validation and Row Level Security (RLS) in its security checkpoint for the software being built, which mitigates runtime issues in the target application.
- [COMMAND_EXECUTION]: The workflow relies on various Model Context Protocol (MCP) tools to interact with the environment. It utilizes Supabase for SQL execution and migrations, Next.js for error detection, and Playwright for automated UI testing and screenshots. These represent standard developer tool interactions required for the skill's stated purpose.
Audit Metadata