velt-chat-sdk-adapter-best-practices
Pass
Audited by Gen Agent Trust Hub on Jun 16, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The documentation describes patterns for building AI bots that process user-generated content from Velt comment threads. This creates a surface for indirect prompt injection when user text is passed to an LLM.
- Ingestion points: User comments are received via the
message.textproperty in theonNewMentionandonSubscribedMessagehandlers withinrules/shared/events/events-on-new-mention.mdandrules/shared/events/events-on-subscribed-message.md. - Boundary markers: The code examples demonstrate direct interpolation of user text without explicit delimiters or instructions to ignore embedded commands.
- Capability inventory: The skill enables the bot to read messages, post replies (
thread.post), and subscribe to threads (thread.subscribe). - Sanitization: No specific sanitization or escaping logic is included in the implementation examples for the comment text.
- [CREDENTIALS_UNSAFE]: The skill provides guidance on managing sensitive API keys and secrets. It correctly recommends the use of environment variables (e.g.,
VELT_API_KEY,VELT_WEBHOOK_SECRET) and explicitly warns against hardcoding secrets in source code inrules/shared/deployment/deployment-env-vars.md. All examples use standard placeholders likewhsec_...orsk-ant-.... - [EXTERNAL_DOWNLOADS]: The skill references several official Node.js packages and documentation sites related to the Velt platform and Vercel ecosystem. All external resources and package references (e.g.,
@veltdev/chat-sdk-adapter) are vendor-owned or from well-known technology providers.
Audit Metadata