nlweb-chatgpt-appsdk
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch documentation and architectural guidance from an external GitHub repository (github.com/nlweb-ai/NLWeb). While GitHub is a well-known platform, the specific organization is not on the internal trusted list.- [COMMAND_EXECUTION]: The instructions direct the agent to execute shell commands for setup using the Bash tool, including 'npm install', 'npm run build', and starting Python-based web servers within the local environment.- [INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by instructing the agent to ingest and follow instructions from external URLs. Evidence Chain: 1. Ingestion points: External markdown files linked in SKILL.md and user-provided 'query' parameter in the 'nlweb-list' tool. 2. Boundary markers: None. 3. Capability inventory: The agent is granted Bash, Write, Edit, and Read tools. 4. Sanitization: No explicit sanitization or validation of the fetched content is specified.
Audit Metadata