Skip to content

Fable vs Fagan

Fable is an open-source MCP server for ai models and llm tooling. Fagan is an open-source MCP server for developer tools.

  • Fable is best for produce a detailed implementation plan with a strong model, then have a cheaper model carry it out.
  • Fagan is best for break a feature goal into dependency-ordered stories and deliver them through review gates.
What it is
Fable is an open-source MCP server for ai models and llm tooling.
Fagan is an open-source MCP server for developer tools.
Best for
Produce a detailed implementation plan with a strong model, then have a cheaper model carry it out
Break a feature goal into dependency-ordered stories and deliver them through review gates
Kind
MCP server
MCP server
Pricing
Open source
Open source
Platforms
CLI, API, self-hosted, Linux, macOS, Windows
CLI, macOS, Linux, API, self-hosted, Web
Open source
Good for
Produce a detailed implementation plan with a strong model, then have a cheaper model carry it out; Get a coverage-first review of a diff or design before merging; Run a security or concurrency-focused critique of a code change
Break a feature goal into dependency-ordered stories and deliver them through review gates; Route small, well-scoped implementation tasks to a local model to reduce token spend; Run an unattended coding loop on a repository while keeping merges human-gated

What Fable does

  • Executor-ready planning

    The plan tool returns numbered steps with exact paths, signatures, edge cases and acceptance criteria meant to be handed to another model.

  • Coverage-first code critique

    The critique tool lists every finding with severity and confidence, leaving filtering to the caller, with an optional focus area.

  • Adjustable reasoning effort

    Effort can be set from low to max per call, and the ask tool can return a readable summary of reasoning.

  • Explicit refusal handling

    Safety refusals are returned with their category and explanation instead of being silently retried on another model.

  • Per-call cost estimates

    Each response includes an estimated USD cost calculated at Fable's listed per-token rates.

  • Portable stdio transport

    Runs over stdio with JSON-RPC 2.0 so it can be registered with any MCP client that supports local servers.

What Fagan does

  • Model role routing

    Each pipeline role can be assigned to Claude, Ollama, LM Studio, MLX or a LiteLLM-routed vendor through a registry or environment variables.

  • Test-first story execution

    Stories run in isolated git worktrees with tests written before implementation, followed by an independent review pass.

  • Risk-tiered overlord

    High-risk decisions such as irreversible actions, auth or money changes are held for a human in dry-run and gated modes.

  • Merge gate with rebase re-test

    Branches are rebased on master, CI is polled, and the suite is re-run before merging.

  • Plan dashboard and scheduler

    A FastAPI dashboard shows a kanban view of plans and stories, and a scheduler daemon can advance ready work unattended.

  • Stuck-story triage

    Optional auto-triage can escalate a struggling local attempt to Claude, split a story, or hold it for review.

Fable vs Fagan: questions

Is Fable or Fagan cheaper?

Both are open source. Compare current plans on each official site.

What is the difference between Fable and Fagan?

Fable is an open-source MCP server for ai models and llm tooling. Fagan is an open-source MCP server for developer tools.

Keep comparing