AI agents use codex_task_create to create or update resources in Openai — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Openai environment.
The tool creates new tasks within a Codex system (likely OpenAI's code execution/task management feature). This is reversible (tasks can be deleted or modified later), produces side effects limited to task state creation, and poses medium risk if misused by an agent—for example, creating excessive tasks, tasks with malicious payloads, or tasks that consume resources.
From the tool's definition Tool name 'codex_task_create' and description 'Create a new Codex task' indicate data creation/modification. The term 'create' is a classic Write operation that adds new records without permanent destruction or financial impact.
Attacks that exploit this kind of access
Create a new Codex task. It is categorised as a Write tool in the Openai MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Openai MCP server in PolicyLayer and add a rule for codex_task_create: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Openai. Nothing to install.
codex_task_create is a Write tool with medium risk. Write tools should be rate-limited to prevent accidental bulk modifications.
Yes. Add a rate_limit block to the codex_task_create rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for codex_task_create. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
codex_task_create is provided by the Openai MCP server (robotlearning123/gpt2agent). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Every MCP server has a record like this.
Type a name, get the same breakdown: verified identity, auth posture, risk grade, capabilities, recommended policy.
Teams ship this data inside their own products. See what a licence covers →