apideck-mcp
Warn
Audited by Gen Agent Trust Hub on Apr 27, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Instructions direct users to run the
@apideck/mcppackage usingnpx -y @apideck/mcp, which executes the vendor's command-line tool on the local system. - [EXTERNAL_DOWNLOADS]: The skill references and fetches resources from
mcp.apideck.devand the vendor's GitHub repository, both of which are official infrastructure for Apideck. - [REMOTE_CODE_EXECUTION]: Documentation describes an experimental 'Code mode' (
?mode=code) where an agent can execute JavaScript. The skill notes the implementation usesvm.createContextand warns that it is not production-hardened. - [PROMPT_INJECTION]: The skill establishes an ingestion surface for indirect prompt injection through its integration with 330+ external SaaS connectors.
- Ingestion points: Data returned from external SaaS platforms via the
execute_tooloperation. - Boundary markers: The instructions do not define specific delimiters or warnings to separate tool output from agent instructions.
- Capability inventory: The skill enables broad write access to sensitive SaaS domains like Accounting, HRIS, and CRM.
- Sanitization: No explicit validation or filtering of external data is specified before it is incorporated into the agent's context.
Audit Metadata