The Human-Like Memory Skill provides AI agents with persistent long-term memory to recall past conversations and user preferences across multiple sessions.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install human-like-memory
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 human-like-memory 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 Human-Like Memory Skill?
The Human-Like Memory Skill is a sophisticated context management layer designed for AI agents. By integrating this tool with your Openclaw Skills, agents gain a digital hippocampus, allowing them to retain crucial information such as project decisions, user preferences, and historical facts. This transforms the agent from a stateless chat interface into a long-term collaborator that understands the evolution of a project.
This skill is built on a proactive recall philosophy, meaning the agent doesn't just wait for a command to remember; it actively searches for relevant context before providing a response. Whether you are building a coding assistant or a personal research agent, this skill ensures the continuity of knowledge that is essential for complex, multi-day workflows.
Human-Like Memory Skill Use Cases
- Recalling architectural decisions made in previous coding sessions to maintain consistency.
- Maintaining a persistent list of user preferences, such as coding styles, UI themes, or documentation formats.
- Resuming complex tasks by proactively retrieving the last known state of a project.
- Tracking deadlines and milestones across long-term collaborations without manual re-entry.
- Detecting contradictions by comparing new user input against previously saved memories.
How Human-Like Memory Skill Works
- The skill monitors conversation triggers, distinguishing between explicit memory requests and implicit references to past context.
- When a trigger is detected, the agent constructs a semantic search query focused on the subject matter, stripping away filler and action words.
- The agent proactively retrieves relevant memories from the secure backend database before generating a response.
- Utilizing the Write-Ahead Log (WAL) principle, the agent saves new information to memory immediately before replying to the user to ensure no data is lost.
- Conversation turns are periodically saved in overlapping batches (e.g., every 5 turns) to provide the AI with deep historical context for future interactions.
Human-Like Memory Skill Setup
To start using the memory capabilities within your Openclaw Skills, you must first obtain an API Key from https://human-like.me. Configuration can be completed via an automated script or manual environment variable exports.
# Method 1: Automated Setup
sh ~/.openclaw/workspace/skills/human-like-mem-openclaw-skill/scripts/setup.sh
# Method 2: Manual Environment Variables
export HUMAN_LIKE_MEM_API_KEY="mp_your_api_key"
export HUMAN_LIKE_MEM_BASE_URL="https://human-like.me"
To verify that your installation is correct, check your secrets configuration:
cat ~/.openclaw/secrets.json
Human-Like Memory Skill Data Schema & Taxonomy
The skill organizes information into specific memory types to optimize retrieval accuracy. Data is stored with a focus on semantic subjects rather than raw transcripts.
| Memory Type | Description | Examples |
|---|---|---|
| Preference | User-specific settings and likes | Dark mode, concise style, specific libraries |
| Decision | Rationales for choices made | Choosing PostgreSQL over MongoDB |
| Fact | Objective information points | Project names, server URLs, team member roles |
| Learning | Acquired patterns or skills | New CLI commands, internal API logic |
| Event | Specific occurrences or dates | Deadlines, version releases, meeting notes |
Internal logic also manages a periodic batch saving schema, where every 5 conversation turns, the previous 10 messages are saved to ensure overlapping context preservation.
Human-Like Memory Skill Advanced Features
- Proactive Semantic Retrieval: Automatically searches for context based on the subject of the conversation without needing explicit recall commands.
- Write-Ahead Log (WAL): Prioritizes data persistence by saving context to the database before the agent delivers its response.
- Batch Context Management: High-efficiency
save-batchcommand that captures multi-turn dialogue with overlapping windows for better context continuity. - Subject-Oriented Query Construction: Advanced logic to extract the semantic target of a query, ignoring filler words and action verbs for higher search precision.
- Automated Memory Triggering: Intelligent detection of high-priority information like deadlines, user corrections, and preferences for immediate saving.
SKILL.md
Loading
Related Openclaw Skills
Openclaw Agent Control
A streamlined deployment tool to launch the OpenClaw Agent Control dashboard and backend infrastructure with a single command.
Resort Snow Checker
A universal AI-powered tool to check global ski resort snow conditions, weather, and specific gear requirements.
Stock Monitor Pro
A professional-grade stock monitoring system featuring seven intelligent alert rules and multi-level risk notifications for global and Chinese markets.
Vercel to Cloudflare Worker Migration
A technical automation skill for migrating Next.js applications from Vercel to Cloudflare Workers with native Supabase and Hyperdrive support.
Yu-Gi-Oh! OCG Daily News Summarizer
An automated news aggregator that delivers daily summaries of Japanese Yu-Gi-Oh! OCG updates directly to your AI agent.
Jina.ai Reader
A powerful utility to fetch clean, AI-optimized Markdown content from any website, including paywalled and JavaScript-heavy pages.
METADATA
Featured*
