A CLI and workflow engine that runs coding agents like Claude Code, Codex and Copilot through configurable review, repair and test gates before work is accepted.
DannyMac180/fable-advisor
A Claude Code plugin where your session acts as architect, routing implementation to GPT-6 Codex lanes and reviewing finished work with Fable 5.1.
- Stars
- 1,139
- Forks
- 96
- Open issues
- 17
- Last commit
- 24 days ago
What is it?
Fable Advisor is a Claude Code plugin built around an architect pattern. The session model owns requirements, task decomposition, specs and verification. Implementation tasks are routed to one of two lanes: GPT-6 Luna for well-specified routine work, or GPT-6 Sol for harder tasks such as subtle concurrency, debugging and security-sensitive changes. Both lanes run through the OpenAI Codex CLI.
A read-only reviewer agent pinned to Fable 5.1 checks the accumulated changes in a clean context before the architect reports a deliverable as done. It is also consulted at architecture decisions, migrations and API design points. The reviewer never writes code.
The plugin includes a routing skill, a six-part spec format, and a setup command that saves per-lane model and effort choices. A lighter option is a single agent file that adds advisor consults to any Claude Code session.
Key features
- 01Routes implementation tasks to GPT-6 Luna or GPT-6 Sol via the Codex CLI.
- 02Uses a Fable 5.1 read-only reviewer for architecture consults and a final review.
- 03Sets a default reasoning effort per lane, overridable in each spec.
- 04Labels each delegation with the model and effort running it.
- 05Provides a /fable-advisor:setup command to choose lane models.
Use cases
- Delegating routine feature work to a cheaper cross-vendor implementer while keeping architecture in a stronger session model.
- Getting an independent review of a finished change before merging.
- Adding a Claude-based second opinion to an existing Claude Code workflow without the full orchestration layer.
Quick start
- 1
Add the plugin marketplace
claude plugin marketplace add DannyMac180/fable-advisor - 2
Install the plugin
claude plugin install fable-advisor@fable-advisor - 3
Install the Codex CLI for the implementer lanes (requires authentication afterwards)
npm i -g @openai/codex - 4
Run setup once to choose lane models
/fable-advisor:setup
Things to know
- Requires Claude Code 2.1.170 or later and a subscription that includes Fable 5.1 for the default reviewer.
- Both implementation lanes need the OpenAI Codex CLI installed and authenticated, plus access to GPT-6 models.
- The architect and reviewer rely on Claude models, so the full pattern is tied to Claude Code and does not run on claude.ai.
- If a pinned Claude model is unavailable on your account, Claude Code silently falls back to the session model.
Topics
FAQ
- What is fable-advisor?
- A Claude Code plugin where your session acts as architect, routing implementation to GPT-6 Codex lanes and reviewing finished work with Fable 5.1. Fable Advisor is a Claude Code plugin built around an architect pattern. The session model owns requirements, task decomposition, specs and verification. Implementation tasks are routed to one of two lanes: GPT-6 Luna for well-specified routine work, or GPT-6 Sol for harder tasks such as subtle concurrency, debugging and security-sensitive changes. Both lanes run through the OpenAI Codex CLI.
- Is fable-advisor free to use? What license does it have?
- fable-advisor is open source under the MIT license. Check the license terms before using it in commercial products.
- Is fable-advisor still maintained?
- Actively maintained. The last commit to fable-advisor was on Sep 25, 2026.
- What can I use fable-advisor for?
- Delegating routine feature work to a cheaper cross-vendor implementer while keeping architecture in a stronger session model. Getting an independent review of a finished change before merging. Adding a Claude-based second opinion to an existing Claude Code workflow without the full orchestration layer.
- How do I install fable-advisor?
- Add the plugin marketplace: claude plugin marketplace add DannyMac180/fable-advisor Install the plugin: claude plugin install fable-advisor@fable-advisor Install the Codex CLI for the implementer lanes (requires authentication afterwards): npm i -g @openai/codex Run setup once to choose lane models: /fable-advisor:setup
- Where can I get fable-advisor?
- The source code and installation instructions are on GitHub: https://github.com/DannyMac180/fable-advisor
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A local browser-based review surface that lets you annotate agent plans, markdown, HTML and code diffs, then send feedback back to coding agents.
A skills framework and software development methodology that guides coding agents through brainstorming, planning, TDD, and reviewed subagent-driven implementation.
A collection of agent skills for coding assistants like Claude Code and Codex, covering grilling sessions, TDD, bug diagnosis, planning and code review.
A prompt-based skill that makes AI coding agents write only the code a task needs, favoring reuse and minimal solutions over large implementations.
Turns a codebase, its docs, schemas and PDFs into a queryable knowledge graph for AI coding assistants, built locally with tree-sitter AST parsing.