mechinterp-next-step-planner
CommunityPlan the next-step mechinterp experiments.
Education & Research#planning#experiments#hypotheses#mechanistic-interpretability#ExperimentSpec#research-state
Authorcesaregarza
Version1.0.0
Installs0
System Documentation
What problem does it solve?
This Skill analyzes current hypotheses and evidence in mechanistic interpretability research to identify the most informative next experiments and generate ready-to-run specifications.
Core Features & Use Cases
- Analyze hypotheses, evidence, and gaps to recommend 2-3 experiments that maximize discriminative power.
- Generate ExperimentSpec JSON files directly to the specs directory for execution by mechinterp-runner.
- Enforce guardrails (one-rung-per-family, ReLU floor, evidence diversification, hypothesis coverage, token-influence planning) to ensure safe, interpretable experimentation.
Quick Start
Run the planner on the current research state to produce the next-step experiment specs in the specs directory.
Dependency Matrix
Required Modules
None requiredComponents
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: mechinterp-next-step-planner Download link: https://github.com/cesaregarza/SplatNLP/archive/main.zip#mechinterp-next-step-planner Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
Agent Skills Search Helper
Install a tiny helper to your Agent, search and equip skill from 223,000+ vetted skills library on demand.