hypotheses

Review, promote, reject, or edit LLM-inferred behavioral hypotheses.

Updated Apr 10, 2026

Search Tags:#LLM #memory management #reflection #behavioral patterns #hypotheses #behavioral

Ready to run

npx skills add https://github.com/emmahyde/memesis --skill hypotheses

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: hypotheses
Source: https://github.com/emmahyde/memesis/tree/main/skills/hypotheses
Command: npx skills add https://github.com/emmahyde/memesis --skill hypotheses

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill enables users to review, promote, reject, or edit LLM-inferred behavioral hypotheses, streamlining hypothesis management.

Core Features & Use Cases

List hypotheses: Display hypotheses with evidence and session data for review.
Promote hypotheses: Advance validated hypotheses into the stable memory stage for future retrieval.
Reject hypotheses: Archive or dismiss hypotheses that are invalid or incorrect.
Edit hypotheses: Modify existing hypothesis data to correct or update information.
Use Case: A user can list all pending hypotheses, promote those with sufficient evidence, and reject spurious ones to maintain a clean hypothesis space.

Quick Start

Use the hypotheses skill to list all current hypotheses and promote a specific hypothesis by its ID.

Frequently Asked Questions about hypotheses

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I manage LLM-inferred behavioral hypotheses during model validation?▼

Managing LLM-inferred behavioral hypotheses involves reviewing, promoting, rejecting, or editing them to streamline the hypothesis lifecycle. This maintains a clean hypothesis space by advancing validated items to stable memory and dismissing incorrect ones.

What is the best way to promote validated behavioral patterns into stable memory?▼

The best way to promote validated behavioral patterns into stable memory is to advance hypotheses with sufficient evidence by their specific ID. This updates the model's state and ensures validated patterns are available for future retrieval.

Do I need memory and reflection modules to review LLM behavioral hypotheses?▼

Yes, managing LLM behavioral hypotheses requires access to memory and reflection modules. These modules are necessary for the system to execute decision-making and accurately perform state updates during hypothesis lifecycle operations.

Can I edit or update existing behavioral hypotheses to correct invalid data?▼

Yes, you can edit existing behavioral hypotheses to modify and correct invalid or outdated information. This editing function allows you to update hypothesis data directly, ensuring accuracy before promoting or rejecting the entry.

How do I list all pending behavioral hypotheses with their supporting evidence?▼

You can list all pending behavioral hypotheses to display them alongside their corresponding evidence and session data. This review process allows you to evaluate which behavioral patterns have sufficient support for promotion to stable memory.

When should I reject LLM-inferred behavioral hypotheses during analysis?▼

You should reject LLM-inferred behavioral hypotheses when they are invalid, incorrect, or spurious. Rejecting these entries archives or dismisses them from the active queue, which maintains a clean and accurate hypothesis space.