automation-workflows

Warn

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill enables the dynamic execution of custom JavaScript snippets (e.g., return data.map(...)) for data transformation, which are defined within workflow configuration files.
  • [COMMAND_EXECUTION]: The skill utilizes a custom CLI tool (claw) for workflow management and suggests the use of system cron for task scheduling, which involves modifying system-level task configurations.
  • [EXTERNAL_DOWNLOADS]: The skill communicates with the vendor's API at api.openclaw.com and supports the configuration of outbound webhooks to interact with external services.
  • [PROMPT_INJECTION]: The skill identifies an attack surface for indirect prompt injection by processing external data from untrusted sources.
  • Ingestion points: Untrusted data enters the agent context via HTTP POST triggers and webhook integrations (e.g., GitHub events) described in SKILL.md.
  • Boundary markers: No specific boundary markers or 'ignore' instructions are defined to isolate untrusted payload data from the processing logic.
  • Capability inventory: The skill is capable of executing JavaScript snippets, invoking CLI commands via the claw tool, and performing network operations using the Python requests library.
  • Sanitization: There is no documentation regarding the sanitization or validation of ingested payloads before they are passed to transformation functions or outbound webhooks.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 7, 2026, 05:44 PM