setup-methods

Official

Scaffold reusable interpretability methods

Authorgoodfire-ai
Version1.0.0
Installs0

System Documentation

What problem does it solve?

It removes the friction of creating new session-local interpretability “method” primitives by turning a structured method spec into runnable Python code with tests and documentation, following strict layering rules.

Core Features & Use Cases

  • Scaffold session-local method code: Generates ${SESSION_DIR}/code/methods/<name>/ with an importable module, a placeholder implementation, and a smoke test.
  • Elicit or consume method specifications: Supports interactive single-spec elicitation or batch scaffolding from one or more provided spec paths.
  • Enforce interpretability-layer invariants: Refuses forbidden imports (e.g., from runner/ or analyses/) and keeps methods pure (no disk I/O and no hyperparameter defaults in code).
  • Batch approval workflow: In interactive mode, presents a consolidated batch summary so you can approve or edit specs before code scaffolding.

Quick Start

Run /research-session first, then invoke setup-methods with one or more method spec paths, for example: “/setup-methods path/to/method_1/set_up_method.md path/to/method_2/set_up_method.md”.

Dependency Matrix

Required Modules

None required

Components

Standard package

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: setup-methods
Download link: https://github.com/goodfire-ai/causalab/archive/main.zip#setup-methods

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.