A tagged long-term memory system that stores user preferences and project decisions with built-in human verification for high accuracy.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install tag-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 tag-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 TagMemory?
TagMemory is a specialized long-term memory layer for AI agents, allowing them to persistently store and retrieve critical user information. By utilizing a semantic tagging system and BM25 search algorithms, this skill ensures that agents powered by Openclaw Skills can remember specific user preferences, architectural decisions, and project contexts across different sessions. The system is built on a privacy-first philosophy, keeping all data stored locally.
What sets TagMemory apart is its focus on data integrity through human-in-the-loop verification. It doesn't just store information blindly; it includes workflows for periodic summarization and manual audits, ensuring the AI's internal knowledge remains accurate and relevant over time. This makes it an essential tool for developers who need their agents to maintain a high level of situational awareness without the volatility of standard context windows.
TagMemory Use Cases
- Recording coding style preferences like tab indentation or specific library versions to ensure consistency in generated code.
- Storing the rationale behind major project decisions, such as choosing one database over another, for future reference.
- Tracking project-specific metadata, including team roles, stakeholder names, and historical event logs.
- Summarizing weekly progress and discussions to maintain a high-level overview of project evolution.
- Correcting or refining AI-stored memories through a proactive verification system to prevent knowledge drift.
How TagMemory Works
- Storage: When a user provides important info, the skill captures the text, assigns semantic tags (like #preference or #decision), and records the source agent ID.
- Indexing: The content is indexed using BM25, a state-of-the-art keyword ranking function, to allow for precise retrieval during future queries.
- Retrieval: When asked about past events or preferences, the skill searches the local database using keywords and tags to return the most relevant records.
- Verification: The system flags new or uncertain memories for human review, allowing users to confirm, edit, or delete them during idle time.
- Archiving: Periodic summaries are generated from recent memories; once confirmed by the user, these summaries are archived as consolidated knowledge.
TagMemory Setup
To get started with TagMemory in your Openclaw Skills environment, follow these steps:
Navigate to the skill directory:
cd ~/.openclaw/workspace/skills/tag-memory
Store a new memory via the CLI:
python3 src/cli.py --json store << EOF
{"content": "User prefers PostgreSQL for all production workloads", "tags": ["#preference", "#database"], "agent_id": "main"}
EOF
Query existing memories to test retrieval:
python3 src/cli.py --json query << EOF
{"query": "database preference", "tags": ["#preference"]}
EOF
TagMemory Data Schema & Taxonomy
TagMemory organizes data locally to ensure privacy and speed. The schema focuses on semantic clarity and traceability.
| Field | Description | Type |
|---|---|---|
content |
The actual text of the memory | String |
tags |
Array of semantic markers (e.g., #decision, #person) | Array |
agent_id |
ID of the specific agent that stored the memory | String |
timestamp |
ISO format date of creation | DateTime |
status |
Verification status (pending/verified) | String |
Common Tags:
#preference: User habits and coding styles.#decision: Important project-level choices.#project: General context regarding the current workspace.#knowledge: Specific facts or techniques learned during interaction.
TagMemory Advanced Features
- Multi-Agent Support: Track which specific agent generated each memory using the agent_id field, perfect for multi-agent Openclaw Skills workflows.
- BM25 Relevance Ranking: High-performance keyword search that outperforms simple string matching for large memory sets.
- Proactive Audit System: Commands like
verify-pendingallow the agent to clean up its own database by asking the user for confirmation. - Time-Aware Queries: Retrieve memories from specific date ranges to reconstruct project timelines.
- Local-Only Storage: Ensures sensitive project data never leaves your local machine, maintaining strict privacy standards.
SKILL.md
Loading
Related Openclaw Skills
Multi Search Engine
A versatile search integration providing access to 10 major global and regional search engines for comprehensive information retrieval.
Geely Knowledge Card AI Skill
A high-efficiency AI skill designed to provide instant, structured knowledge synthesis and strategic business facts about Geely.
Louis Vuitton
A specialized AI assistant designed to retrieve real-time product, financial, and corporate data for Louis Vuitton.
LEGO Encyclopedia Guide
A comprehensive digital guide for LEGO enthusiasts to explore sets, themes, history, and expert building advice.
ICCircle News
An AI agent skill that fetches and aggregates semiconductor industry news and VLSI blog updates from multiple global sources.
Tavily arXiv Paper Fetech
An automated pipeline that resolves scientific paper titles into verified arXiv metadata and generates structured research reports.
Featured*
