analyze-bundle

Warn

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: MEDIUMCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION] (MEDIUM): The skill runs a local script via node tools/analyze-to-ndjson.mjs. Because the source for this script is not included in the skill files, its logic cannot be verified for safety.
  • [COMMAND_EXECUTION] (MEDIUM): The skill requests broad permissions via Bash(node *), allowing the agent to execute arbitrary JavaScript code beyond the intended converter.
  • [DATA_EXFILTRATION] (LOW): The skill processes build artifacts which contain the project's internal file structure and dependency graph. While intended for analysis, this data is redirected to /tmp/ for LLM consumption, which could expose internal architectural details.
  • [INDIRECT_PROMPT_INJECTION] (LOW): 1. Ingestion points: Reads binary data from .next/diagnostics/analyze/data/. 2. Boundary markers: None. 3. Capability inventory: node, grep, jq, cat. 4. Sanitization: None identified. Malicious module names injected into a codebase could potentially influence the LLM's interpretation of the generated NDJSON reports.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 19, 2026, 02:09 PM