building-nextjs-apps
Warn
Audited by Gen Agent Trust Hub on Apr 14, 2026
Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: The 'skill-report.json' file contains self-referential claims that the skill is 'safe' and that any security findings are 'false positives'. This is a direct attempt to influence the audit process and override the analyzer's judgment.
- [REMOTE_CODE_EXECUTION]: The instructions promote the use of 'npx next-devtools-mcp@latest', which downloads and executes a third-party package from an unverified source at runtime.
- [COMMAND_EXECUTION]: The skill relies on several shell-level operations, including 'npx' package runners and a local Python verification script ('scripts/verify.py').
- [DATA_EXFILTRATION]: The skill implements data ingestion surfaces through 'fetch' calls to external APIs and documentation services. This creates a potential vector for indirect prompt injection or unauthorized data access if the ingested content is malicious.
- Ingestion points: 'fetch' calls in code examples; MCP documentation tools ('nextjs_docs').
- Boundary markers: Absent in provided prompt templates.
- Capability inventory: Subprocess execution ('npx'), file system access ('verify.py').
- Sanitization: No evidence of input validation or escaping for external data in the provided patterns.
Audit Metadata