LaunchKit · 2026
Back to Skills

turing-pyramid

Decision framework for agent psychological health. 10 needs with decay, tension-based priority, cross-need cascades.

2
877 downloads
by @tensusds

Setup & Installation

openclaw skills install @tensusds/turing-pyramid

Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:

npx clawhub install turing-pyramid

Version History

v1.22.1Mar 16, 2026 - Version note by skill developer:

Spontaneity tuning: gate 1.5->1.0, baseline 2.0->1.5, threshold 10->6. First spontaneous event in ~16-24h instead of ~80h

What This Skill Does

Prioritized action selection engine for AI agents using 10 configurable needs with time-decay and tension scoring. Each cycle calculates which needs are most deprived and selects concrete actions ranked by urgency. Replaces idle heartbeat loops with a decision framework that adapts to agent state over time.

Instead of polling for tasks or waiting for user input, agents get a ranked action list derived from measurable deprivation levels, decay curves, and cross-need cascades.

When to use it

  • Running an AI agent on a regular heartbeat schedule with meaningful next steps
  • Preventing an autonomous agent from cycling without purpose between tasks
  • Tuning an agent's priorities for research-focused versus social-facing work
  • Auditing what actions an agent took and why over the past week
  • Detecting when a low-priority need has been neglected for too long

Example Workflow

Here's how your AI assistant might use this skill in practice.

INPUT

User asks: Run the Turing Pyramid cycle for my agent

AGENT
  1. 1Reads current satisfaction values from assets/needs-state.json
  2. 2Applies time-decay to each need based on hours elapsed and the configured day/night multiplier
  3. 3Calculates tension scores (importance × (3 - satisfaction)) for all 10 needs
  4. 4Selects highest-tension needs using probability tables and impact range matrices
  5. 5Outputs ranked action suggestions with mark-satisfied.sh commands for each
OUTPUT

A prioritized list of 2-3 concrete actions with tension scores, impact values, and follow-up mark-satisfied commands

Requirements

Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.

WORKSPACE environment variable set to the agent workspace directory path