run-parallel-agents-feature-debug
Warn
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill's core functionality involves the dynamic construction of prompts for the Agent tool. In Step 3, the coordinator assembles instructions including reproduction steps, error symptoms, and scope boundaries based on the current project state and user input. This runtime assembly of instructions for spawned agents constitutes dynamic execution of generated logic.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a significant attack surface for indirect prompt injection. It is designed to ingest data from potentially untrusted external sources, such as stack traces, test output, and file contents, using Read, Glob, and Grep tools. 1. Ingestion points: External data is ingested in Step 3 ('Build full debugging briefs') from files and command outputs. 2. Boundary markers: The skill instructions do not specify the use of delimiters (e.g., XML tags or triple quotes) or 'ignore embedded instructions' warnings when passing this data to sub-agents. 3. Capability inventory: The skill and its sub-agents have access to powerful tools including Agent (for further spawning), Bash (shell execution), and filesystem tools (Read, Glob, Grep). 4. Sanitization: There is no mention of sanitization, validation, or escaping of the ingested data before it is interpolated into the prompts for sub-agents.
Audit Metadata