chat-sdk

Pass

Audited by Gen Agent Trust Hub on Jul 23, 2026

Risk Level: SAFE
Full Analysis
  • Indirect Prompt Injection Surface: The skill defines handlers that process incoming messages from external chat platforms such as Slack, Discord, and Telegram. This creates a surface where external content could potentially influence the agent's behavior if not properly handled.
  • Ingestion Points: The skill uses handlers like onNewMention and onSubscribedMessage in SKILL.md to receive user-provided message text.
  • Capability Inventory: The SDK provides capabilities to post messages and stream content back to these platforms using thread.post() and thread.stream().
  • Sanitization: While the SDK likely includes its own normalization, developers are encouraged to review the referenced documentation in node_modules/chat/docs/handling-events.mdx for best practices on processing external input.
  • Boundary Markers: No explicit prompt boundary markers are demonstrated in the high-level examples.
  • External References: The skill references the chat npm package and several adapter packages (e.g., @chat-adapter/slack, @chat-adapter/state-redis). It also includes a link to https://chat-sdk.dev/adapters for additional resources. These references are used to facilitate the intended development workflow.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 23, 2026, 10:11 AM
Security Audit — agent-trust-hub — chat-sdk