A sophisticated memory management system that organizes scattered agent logs and assets into a structured, layered knowledge runtime with typed relationships.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install openclaw-knowledge-runtime
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 openclaw-knowledge-runtime 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 Openclaw Knowledge Runtime?
The Openclaw Knowledge Runtime is a specialized framework designed to transform fragmented agent data into a high-performance memory layer. It enables AI agents to maintain a long-term, organized understanding of their environment by categorizing information into distinct layers such as episodic, semantic, and procedural. By implementing this within the ecosystem of Openclaw Skills, developers can ensure their agents move beyond simple logging to sophisticated knowledge synthesis.
This skill focuses on the structural integrity of AI memory, providing a blueprint for linking knowledge entries to specific entities, genes, and historical tasks. It effectively solves the problem of noisy memory by enforcing strict write-back rules, ensuring only stable and validated findings are persisted for future runs, thereby increasing the reliability and intelligence of the agent over time.
Openclaw Knowledge Runtime Use Cases
- Consolidating scattered memory logs, tasks, and assets into a unified runtime.
- Implementing typed relationships between knowledge entries, entities, and events for better context.
- Generating compact, high-relevance retrieval bundles for LLM prompts and task ranking.
- Establishing a durable research store that prevents speculative or temporary data from polluting long-term memory.
- Enhancing agent observability through detailed tracking of hit counts and layer coverage.
How Openclaw Knowledge Runtime Works
- Define a two-axis memory model using specific layers (working, episodic, semantic, procedural, policy) and scopes (session, shared, published).
- Initialize storage for core records including knowledge entries, typed links, and canonical entities.
- Construct dynamic queries based on the agent's current role, objective, and real-time signals.
- Execute retrieval by ranking candidates and expanding context through one-hop typed link analysis.
- Generate a compact output bundle containing hits, bias tags, and linked assets for the agent's prompt.
- Perform a strict write-back phase to record only high-signal, validated findings after a successful run execution.
Openclaw Knowledge Runtime Setup
To integrate the knowledge runtime into your agent architecture, establish the directory structure and initialize the core store files.
# Create the knowledge storage directory
mkdir -p memory/knowledge
# Initialize the core storage and index files
touch memory/knowledge/knowledge_store.jsonl
touch memory/knowledge/knowledge_links.jsonl
echo '{}' > memory/knowledge/knowledge_index.json
echo '{}' > memory/knowledge/entity_index.json
Ensure your agent environment has access to these paths to utilize the Openclaw Skills memory management features.
Openclaw Knowledge Runtime Data Schema & Taxonomy
The Openclaw Knowledge Runtime organizes data into a structured taxonomy to ensure quick retrieval and durable storage:
| Record Type | Description |
|---|---|
knowledge_entry |
The primary unit of stored knowledge or findings. |
knowledge_link |
A typed relationship (e.g., supports_gene, contradicts) between records. |
entity |
Canonical forms of modules, repos, topics, or organizations. |
Memory Layers:
- Procedural: Skills, capsules, and reusable playbooks.
- Episodic: Event logs and run histories.
- Semantic: Research briefs and stable conclusions.
- Policy: System rules and user-defined constraints.
Openclaw Knowledge Runtime Advanced Features
- Multi-layer retrieval logic that prioritizes policy and procedural knowledge for system stability.
- One-hop link expansion to automatically discover related entities and contradictory evidence.
- Decoupled adapter surfaces allowing the runtime to plug into any agent loop via generic query builders and task rankers.
- Intelligent context trimming to ensure prompt windows are populated only with the highest-signal data.
- Durable write-back filters that distinguish between speculative 'scratchpad' notes and validated research.
SKILL.md
Loading
Related Openclaw Skills
xnyjob_search
A professional job search skill tailored for the new energy sector, allowing users to query positions by keyword, location, and industry.
Product Opportunity Research
A multi-agent framework that utilizes six specialized AI personas to conduct deep product opportunity research and generate actionable business roadmaps.
Magyar Szövegfeldolgozás
A specialized toolkit for processing Hungarian text through AI-driven summarization, keyword extraction, and entity recognition.
ProductivityHub Design Skill
An AI agent skill that integrates ProductivityHub tools and evidence-based guides to automate task prioritization, habit tracking, and time management.
Sidebar Generator
A strategic design framework for building and auditing high-performance sidebars for blogs, documentation, and content-heavy websites.
Bank List Query Skill
An automated directory skill for querying supported banks, integration statuses, and functional capabilities within financial management systems.
Featured*
