A high-performance semantic search runtime designed for real-time conversational AI and on-device agent retrieval with sub-10ms latency.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install moss
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 moss 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 Moss Agent Skills?
Moss serves as a real-time semantic search runtime specifically engineered for conversational AI agents. It provides a robust framework for delivering instant lookups and index updates that can run seamlessly in the browser, on-device, or in the cloud. By integrating this capability with other Openclaw Skills, developers can empower their agents to manage document lifecycles, create specialized indexes, and perform hybrid searches without the burden of managing complex backend infrastructure.
The platform is unique in its ability to handle embedding generation and index persistence internally. This allow agents to focus entirely on retrieval logic and context grounding. Whether you are building a speed-optimized edge application or a high-accuracy cloud-based assistant, these Openclaw Skills provide the necessary tools for lightning-fast data retrieval and knowledge management.
Moss Agent Skills Use Cases
- Building RAG-enabled conversational agents that require sub-10ms response times for fluid interactions.
- Implementing offline-first semantic search for mobile or edge devices where connectivity is limited.
- Enhancing voice agents with real-time context injection using frameworks like LiveKit or Pipecat.
- Managing dynamic document lifecycles with instant index updates without requiring a full reindexing of the corpus.
- Utilizing Openclaw Skills to provide grounded, context-aware answers in AI-driven customer support tools.
How Moss Agent Skills Works
- Initialize the MossClient using your specific MOSS_PROJECT_ID and MOSS_PROJECT_KEY credentials.
- Create a new semantic index by providing a document set and selecting an embedding model such as moss-minilm.
- Load the index from persistent storage into the runtime environment to prepare for active querying.
- Perform semantic or hybrid searches using natural language queries to retrieve the top-K matching documents.
- Optionally update the index in real-time by adding or deleting documents as the agent's knowledge base evolves.
Moss Agent Skills Setup
To get started with these Openclaw Skills, install the appropriate package for your development environment.
For JavaScript/TypeScript:
npm install @inferedge/moss
For Python:
pip install inferedge-moss
Configure your environment variables to authenticate the agent:
export MOSS_PROJECT_ID=your_project_id
export MOSS_PROJECT_KEY=your_project_key
Moss Agent Skills Data Schema & Taxonomy
Moss organizes data through a structured document schema and index metadata. The following table describes the document structure used within Openclaw Skills:
| Field | Type | Required | Description |
|---|---|---|---|
| id | string | Yes | Unique identifier for the document (used for upserts) |
| text | string | Yes | The actual content to be embedded and searched |
| metadata | object | No | Custom key-value pairs for metadata filtering |
Indexes also track metadata such as document count, dimension size, and the specific model ID used (e.g., moss-minilm or moss-mediumlm).
Moss Agent Skills Advanced Features
- Hybrid Search Blending: Combine semantic vector similarity with BM25 keyword matching using configurable alpha weighting.
- Metadata Filtering: Constrain search results based on document categories, languages, or custom tags for precise retrieval.
- Edge-First Runtime: Execute queries entirely on-device or in-browser to eliminate network latency and improve privacy.
- Voice Agent Integration: Specialized support for Pipecat and LiveKit pipelines for automatic context injection.
- Instant Index Updates: Perform document additions and deletions that are immediately reflected in search results without infrastructure delays, a core benefit of these Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
Unified Search Skill
A smart, multi-engine search skill that automatically routes queries to Baidu, DuckDuckGo, or CoinGecko based on language and context.
Mermaid Image Generation & Hosting
An automated tool to transform Mermaid diagrams into hosted image URLs for seamless Markdown and content integration.
Draw.io Flowchart Generator
An intelligent generator that creates professional Draw.io system architecture diagrams and flowcharts with automated local app integration.
Brutal Honesty Advisor
An unfiltered AI feedback engine designed to provide high-signal, objective reality checks for ideas and decisions.
Firefly AI Integration
Retrieve meeting transcripts, summaries, and action items directly from fireflies.ai using GraphQL.
AgoraHub
AgoraHub is an open agent registry providing over 14 verified AI tools for common developer tasks like data conversion, hashing, and formatting.
METADATA
Featured*
