chatgpt-task-router
Warn
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONSAFE
Full Analysis
- [DYNAMIC_EXECUTION]: The skill utilizes a dynamic
import()call to load thechatgpt-bridge.mjsscript from a string-defined path. Dynamic loading of modules at runtime can be used to execute code that is not statically verifiable within the skill's primary instruction set. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-controlled data (
userRequest) to determine the routing logic and conversation mode for an external AI service. - Ingestion points: Untrusted data enters the agent context through the
userRequestvariable passed to theclassifyChatGPTConversationModefunction inSKILL.md. - Boundary markers: The skill lacks explicit boundary markers or instructions to the LLM to ignore potentially malicious commands embedded within the
userRequesttext. - Capability inventory: The skill possesses the capability to control an in-app browser (
browser:control-in-app-browser) and interact with an external web interface. - Sanitization: There is no evidence of sanitization or input validation performed on the user-provided text before it is used to influence the agent's decision-making logic.
- [SAFE]: The skill references and interacts with
chatgpt.com. As this is a well-known service provided by OpenAI, the network interaction with this domain is considered standard behavior for the stated purpose of the skill.
Audit Metadata