Cognitive

Manage working memory, episodic memory, and semantic retrieval with confidence estimation.

3|Updated Apr 18, 2026
One-click install
npx skills add https://github.com/garochee33/DSH --skill cognitive-garochee33
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: Cognitive
Source: https://github.com/garochee33/DSH/tree/main/kb/skills/cognitive
Command: npx skills add https://github.com/garochee33/DSH --skill cognitive-garochee33

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires torch, numpy, scipy, chromadb.

What problem does it solve?

This Skill helps agents reason more reliably by managing context, recalling relevant experiences, and estimating confidence when evidence is incomplete or noisy.

Core Features & Use Cases

  • Structured reasoning: Supports chain-of-thought, tree-of-thought, and self-consistency workflows for complex problem solving.
  • Memory management: Maintains working, episodic, and semantic memory to preserve useful context across turns and sessions.
  • Attention and belief updates: Scores relevance, weights salient information, and updates beliefs using Bayesian primitives.
  • Use case: An agent can summarize a long conversation, retrieve related prior facts from memory, and decide how confident it should be before responding.

Quick Start

Use the cognitive skill to summarize the current conversation, retrieve relevant prior memories, and return a confidence estimate for the answer.

Frequently Asked Questions about Cognitive

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

FAQPage Schema
How do I maintain working memory and episodic memory across multi-turn conversations?

Use memory management to maintain working, episodic, and semantic memory, preserving useful context across turns and sessions for consistent multi-turn conversations.

How does Bayesian belief updating improve agent reasoning with incomplete evidence?

Bayesian belief updating improves agent reasoning by scoring relevance, weighting salient information, and estimating confidence when evidence is incomplete or noisy during information retrieval.

Can I use ChromaDB for semantic memory retrieval in long-context summarization?

Yes, ChromaDB-backed retrieval supports semantic memory retrieval, allowing agents to retrieve related prior facts from memory and summarize long conversations effectively.

Do I need PyTorch and NumPy to implement attention modeling and confidence estimation?

Yes, PyTorch and NumPy are required dependencies for attention modeling and confidence estimation, along with SciPy and ChromaDB for structured reasoning primitives and retrieval.

What's the best way to apply chain-of-thought and tree-of-thought workflows for complex problem solving?

Structured reasoning workflows like chain-of-thought, tree-of-thought, and self-consistency support complex problem solving by combining memory management with attention and belief updates.