Human Like Memory

What This Skill Does

The Human-Like Memory skill, developed by jianghaibobo2015-rgb, provides the OpenClaw AI agent with persistent, long-term memory capabilities. Unlike standard sessions that lose context once a chat is cleared, this skill allows the agent to store and retrieve past interactions, user preferences, and project-specific data. By leveraging an external API (human-like.me), the agent creates a searchable database of past turns, enabling it to act as a continuity-focused assistant that remembers past decisions, architectural preferences, and previous task states without needing to be reminded.

Installation

To install this skill, use the ClawHub command line: clawhub install openclaw/skills/skills/jianghaibobo2015-rgb/human-like-memory

Once installed, you must configure your API key obtained from https://human-like.me. You can automate this using sh ~/.openclaw/workspace/skills/human-like-mem-openclaw-skill/scripts/setup.sh or by manually exporting the HUMAN_LIKE_MEM_API_KEY environment variable in your shell profile. Verify the configuration by running node ~/.openclaw/skills/human-like-memory/scripts/memory.mjs config.

Use Cases

• Project Continuity: Maintain state across long-running development projects by recalling previous API design choices or debugging notes.
• Personalized Assistance: Save user preferences regarding coding styles, tone, or specific project constraints so you don't have to repeat them.
• Decision Tracking: Keep a history of why specific architectural decisions were made during brainstorming sessions.
• Contextual Awareness: Improve query accuracy by having the agent search its internal memory before providing responses when the user makes vague references like "the last project" or "my usual config."

Example Prompts

1. "Do you remember the decision we made last week regarding the authentication provider for the dashboard project?"
2. "What are my preferred coding styles that I've mentioned in our previous conversations?"
3. "Recall the API design specifications we discussed yesterday so we can keep working on the implementation."

Tips & Limitations

• Be Proactive: Do not wait for the user to ask. If you suspect the current prompt relates to past data, use the recall command before drafting your response.
• Data Privacy: This skill transmits conversation data to an external API (human-like.me). Ensure your data handling policies align with the privacy terms of the provider.
• Batching: Use save-batch for large imports of existing context to minimize API call overhead.
• Maintenance: Regularly run config to ensure your session token or environment variables haven't expired.