automate-whatsapp
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFECREDENTIALS_UNSAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [CREDENTIALS_UNSAFE]: The skill manages sensitive credentials including platform API keys (
KAPSO_API_KEY) and GitHub Personal Access Tokens (PATs). These are required for authenticating with the Kapso API and for the AI agent sandbox to clone and modify remote repositories. - [COMMAND_EXECUTION]: The "Remote Sandbox" feature allows AI agent nodes to execute shell commands (
bash) and perform file operations (read,write,edit) within an ephemeral workspace. This capability is intended for repository inspection and modification during workflow runs. - [EXTERNAL_DOWNLOADS]: The
scripts/openapi-explore.mjsutility fetches OpenAPI specifications from the vendor's documentation site (docs.kapso.ai). Additionally, the sandbox feature involves cloning external GitHub repositories to perform tasks. - [INDIRECT_PROMPT_INJECTION]: The workflow system is designed to ingest and process untrusted data from WhatsApp messages (via
wait_for_responsenodes). This data is often used as context for AI agent nodes or decision logic, creating a surface for indirect prompt injection attacks. - [DYNAMIC_EXECUTION]: The skill facilitates the creation, deployment, and invocation of custom JavaScript functions on the Kapso platform runtime. This allows for the execution of user-supplied code to handle workflow logic, external API calls, and data processing.
Audit Metadata