A structured long-term memory retrieval engine that transforms messy AI logs into a high-order hierarchical indexing system.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memory-index
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install memory-index using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Memory Index?
Memory Index is a specialized architectural framework designed to solve the problem of context drift and Token waste in AI Agents. Traditionally, AI memories are cluttered in daily log files, leading to hallucinations and cross-task interference. This skill establishes a structured external hippocampus using a three-layer hierarchy—Global Index, Workspace Domains, and Topic Entities—ensuring absolute order and precise memory addressing.
By implementing this Openclaw Skills protocol, developers can enable their agents to reconstruct complex task contexts with minimal Token consumption. It enforces strict safety boundaries and single-source-of-truth principles, making it indispensable for long-term project management and multi-agent collaboration.
Memory Index Use Cases
- Managing long-term software development features that span multiple days or sessions.
- Orchestrating multi-agent handoffs where specific context must be passed without bloating the prompt.
- Reducing Token overhead in complex research or analysis tasks by retrieving only relevant topic metadata.
- Preventing role confusion in single-agent environments by using structured persona-tagged task files.
- Recovering from context loss or system restarts using a centralized active memory index.
How Memory Index Works
- System Initialization: The agent checks for the existence of the active memory index and creates the hierarchical folder structure if missing.
- Task Categorization: New requirements are routed to specific domains (e.g., coding, writing, analysis) based on their technical nature.
- Entity Creation: A dedicated Markdown file is generated for the task, following a strict template that includes roles, goals, and current status.
- Indexing: The task's absolute path and metadata are registered in the central index file (MEMORY_INDEX_ACTIVE.md) for global routing.
- Retrieval & Update: During new sessions, the agent reads the index first, follows the pointer to the specific task file, and updates progress in real-time.
- Distillation: Once completed, tasks are archived and summarized to keep the active context window lean and efficient.
Memory Index Setup
To initialize the Memory Index environment within your project, ensure the following directory structure is created:
mkdir -p memory/topics/coding
mkdir -p memory/topics/writing
mkdir -p memory/topics/analysis
touch memory/MEMORY_INDEX_ACTIVE.md
touch memory/MEMORY_INDEX_ARCHIVED.md
Configure your agent's system prompt to prioritize the MEMORY_INDEX_ACTIVE.md as the primary routing table for all long-term Openclaw Skills operations.
Memory Index Data Schema & Taxonomy
The skill organizes data into a strictly governed three-layer hierarchy:
| Layer | Name | Description |
|---|---|---|
| Layer 1 | System Index | MEMORY_INDEX_ACTIVE.md - The global DNS/Routing table containing metadata and absolute file paths. |
| Layer 2 | Workspace Domains | Physical directories (e.g., /coding/, /analysis/) that prevent file name collisions between different domains. |
| Layer 3 | Topic Entities | Individual Markdown files containing the Single Source of Truth for a specific task, including Milestones and Scratchpads. |
Task Status Taxonomy:
[ACTIVE]: Currently being processed.[PAUSED]: Pending further input or resources.[ARCHIVED]: Completed or discarded; moved to the archive index.
Memory Index Advanced Features
- Cold Start Recovery: Automatic disaster recovery mechanism that uses deep-scan grep to rebuild the index if lost.
- Multi-Agent Handoff: Protocol for passing absolute memory pointers between agents via sessions_spawn to avoid prompt bloat.
- Heartbeat Audit: Periodic maintenance routine that reconciles physical topic files with the central index to clean up 'orphan' files.
- Persona Tagging: Strict prefixing within task files (e.g., [CTO], [PM]) to maintain character consistency across different sub-tasks.
- Token Distillation: Automated archival workflow that strips process logs while retaining core implementation logic to save context space.
SKILL.md
Loading
Related Openclaw Skills
Grafana Inspector
An automated inspection tool for Grafana that combines browser screenshots and API data to generate detailed dashboard reports.
AI Video Editor (Sparki)
An AI-powered video processing skill that automates editing, resizing, and highlight generation for social media platforms.
QC Backtest Master
A streamlined pipeline for automating QuantConnect backtest lifecycles from local strategy files to detailed cloud result downloads.
Backtest Poller
An intelligent background daemon for Openclaw Skills that automates QuantConnect backtest monitoring, equity tracking, and forensic diagnosis.
Opencode
A comprehensive AI workflow engine designed for automated code generation, structural project analysis, and streamlined environment management.
Daily-to-Goal (D2G) MCP Integration
An advanced MCP skill that connects AI agents to the Daily-to-Goal platform for seamless management of organizational goals, tasks, and team metrics.
Featured*
