chat-sdk
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection Surface: The skill is designed to build chat bots that ingest and process messages from external platforms (e.g., Slack, Discord, Telegram). This creates a potential surface where malicious instructions embedded in incoming chat messages could influence the behavior of an integrated AI model.\n
- Ingestion points: The skill utilizes event handlers like
onNewMention,onDirectMessage, andonSubscribedMessageinSKILL.mdto receive data from external chat platforms.\n - Boundary markers: The provided code examples in
SKILL.mddo not explicitly demonstrate the use of delimiters or 'ignore' instructions to isolate untrusted input when passed to AI models.\n - Capability inventory: The SDK facilitates network operations to chat platform APIs and persistence through various state adapters like Redis and PostgreSQL as described in
SKILL.md.\n - Sanitization: Users should implement their own validation and sanitization when interpolating
message.textinto AI prompts to mitigate potential injection risks, as no explicit sanitization is shown in the examples.
Audit Metadata