debug

Pass

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes various shell scripts (metro.sh, logs.sh, hmr.sh) and system utilities like xcrun to manage the local React Native environment. It also provides a way to evaluate arbitrary JavaScript expressions within the application via cdp-bridge.js eval.
  • [DATA_EXFILTRATION]: The skill captures and processes application console logs and network traffic. While intended for debugging, this involves handling runtime data that may contain sensitive information if logged by the application.
  • [PROMPT_INJECTION]: The skill has an indirect prompt injection surface (Category 8).
  • Ingestion points: Untrusted data from the debugged application enters the agent context via cdp-bridge.js console, cdp-bridge.js network, and cdp-bridge.js tree as described in SKILL.md.
  • Boundary markers: Subagent prompts lack explicit delimiters or instructions to ignore potential commands embedded in the captured application data.
  • Capability inventory: The skill is granted Bash, Read, and Agent tools, which could be exploited if an injection occurs.
  • Sanitization: There is no evidence of sanitization or validation of the ingested logs or network data before processing by subagents.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 11, 2026, 10:21 AM