Fable directs the work while Codex builds and researches
This workflow uses Fable as the planner and reviewer, while several Codex agents work in parallel as builders or researchers. The goal is to lower token costs and make coding and research work more efficient. The main idea is role separation. Fable manages the overall direction and checks the results, while Codex handles implementation or research tasks. The setup includes two skills named `/architect` and `/architect-research`. It covers advanced areas such as quality control, token saving, context and memory management, MCP, and multi-agent workflows. The workflow is rated 85/100 in value and is marked as active.
Key points
- Fable acts as the planner and reviewer for the workflow.
- Codex agents do coding and research work in parallel.
- The workflow aims to reduce token costs and improve efficiency.
- It centers on two skills: `/architect` and `/architect-research`.
- It is aimed at advanced users working with MCP, context, memory, and multi-agent workflows.
Quick term guide
- token costs
- Token costs are the fees paid for the text an AI model reads and writes.
- token cost
- The money or usage spent when sending text to an AI model and getting text back.
- implementation
- The step where a plan is turned into working code.
- multi-agent workflows
- Workflows where several AI roles work together on a task.
- multi-agent workflow
- A setup where two or more AI tools each handle different parts of a job and pass results between them
- multi-agent
- A setup where several AI agents each handle a different subtask and work together to complete a larger goal.
- agent workflows
- Step-by-step work patterns where an AI agent handles a task.
- agent workflow
- A set of steps an AI follows automatically to complete a series of tasks in order.