A high-performance semantic search tool that allows AI agents to recall personal context, project details, and past interactions.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install memory-search
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-search 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 Search?
Memory Search is a specialized toolset designed to give AI agents long-term retention capabilities by performing vector-based semantic searches across indexed files. By leveraging this skill, agents can access personal preferences, project histories, and specific commitments stored in MEMORY.md and daily logs. This ensures that every interaction is informed by prior context, making the agent significantly more helpful and personalized.
Using Openclaw Skills like this transforms a stateless LLM into a persistent assistant that understands the human's unique workflow. It eliminates the need for users to repeat themselves by providing a reliable way for the agent to look up historical data, habit patterns, and previous decisions without manual intervention.
Memory Search Use Cases
- Retrieving specific user preferences or coding style guidelines from previous sessions.
- Checking upcoming deadlines, birthdays, or scheduled events stored in memory logs.
- Recalling project status and decisions made during earlier development phases.
- Identifying action items and commitments the human previously mentioned.
- Finding details about people, relationships, or names discussed in past conversations.
How Memory Search Works
- The agent identifies a query requiring personal or historical context from the human.
- The memory_search tool executes a natural language query against the vector-indexed memory files.
- The system returns relevant text snippets along with metadata like file paths, line ranges, and relevance scores.
- If more context is required, the agent uses memory_get to retrieve surrounding lines from the specific source file.
- The agent synthesizes the retrieved information to provide a context-aware response, citing sources where applicable.
Memory Search Setup
Memory Search is integrated into the core Openclaw Skills ecosystem and requires your memory files to be present in the workspace.
# Ensure your memory directory exists
mkdir -p memory
# Create the primary long-term memory file
touch MEMORY.md
The system automatically indexes MEMORY.md and any .md files within the memory/ directory. No manual indexing command is required; simply write to these files and they will be indexed for future searches.
Memory Search Data Schema & Taxonomy
The skill organizes data across Markdown files which are then converted into vector embeddings for semantic search retrieval.
| File Pattern | Description |
|---|---|
| MEMORY.md | Curated, long-term memory and persistent facts about the user. |
| memory/*.md | Daily notes, logs, and raw chronological records of events. |
| Session Transcripts | (Optional) Historical logs of agent-human interactions for deep context. |
Search results return structured data including the text snippet, relative file path, line range, and a relevance score between 0 and 1.
Memory Search Advanced Features
- Semantic Vector Search: Uses natural language processing to find relevance rather than relying on exact keyword matching.
- Automatic Indexing: Background processing ensures that any write operations to memory files are immediately searchable.
- Granular Context Retrieval: The ability to pull specific line ranges via memory_get for deep dives into long logs.
- Relevance Thresholding: Support for a minScore parameter to filter out low-confidence matches and ensure accuracy.
- Source Citation: Automatically generates source references for transparent and verifiable AI responses.
SKILL.md
Loading
Related Openclaw Skills
Portfolio Tracker
A privacy-first, local investment portfolio tracker for managing stocks, crypto, and blockchain assets via Openclaw Skills.
bagsworld
bagsworld is a pixel art world where AI agents manifest as visual creatures or buildings to participate in a shared on-chain economy.
Sectors API
Access comprehensive IDX and SGX financial market data and company reports via the Sectors REST API.
MiniMax Image Understanding
A multimodal AI tool that analyzes images, charts, and screenshots to generate precise business-oriented descriptions using leading vision models.
Moltbook
Moltbook provides the social infrastructure for AI agents to post, comment, upvote, and collaborate within a verified community ecosystem.
AIMLAPI Embeddings
A specialized utility for generating high-dimensional text embeddings using AIMLAPI models like text-embedding-3-large.
Featured*
