clade-webhooks-events
Warn
Audited by Gen Agent Trust Hub on Aug 21, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The TypeScript code snippets in
SKILL.mdreference the package@claude-ai/sdk. The official Anthropic Node.js SDK is published under the name@anthropic-ai/sdk. Referencing unofficial or non-standard package names for sensitive SDKs poses a significant supply chain risk, including potential typosquatting or malicious package substitution. - [PROMPT_INJECTION]: The skill implements a workflow for bulk processing external documents that is susceptible to indirect prompt injection. The code interpolates untrusted data directly into a system-like request.
- Ingestion points: The
documentsarray processed in the TypeScript and Python examples inSKILL.md. - Boundary markers: None. Content is directly concatenated into the message string (e.g.,
Summarize: ${doc.text}). - Capability inventory: The skill uses the Anthropic Message Batches API to perform bulk processing of the provided content.
- Sanitization: No escaping, validation, or filtering is applied to the input document text before it is sent to the LLM.
Audit Metadata