A specialized decision-making skill that determines the optimal sequence, priority, and budget for searching across web, local, vector, and graph data sources.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install source-router
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 source-router 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 Source Router?
Source Router is an essential component within the Openclaw Skills ecosystem designed to solve the complexity of multi-source information retrieval. Instead of blindly searching every available database, this skill analyzes task characteristics, time requirements, and existing information gaps to create a precise search plan. It serves as the 'brain' of the search process, ensuring that AI agents use the most cost-effective and relevant sources first.
By categorizing tasks into types like deep research or breaking news, Source Router dynamically adjusts its strategy. It manages everything from freshness requirements (ensuring data is up-to-date) to counter-evidence searching, which helps eliminate confirmation bias in AI-generated conclusions. This ensures that the agent's research is both comprehensive and logically sound.
Source Router Use Cases
- Orchestrating complex deep research tasks across multiple internal and external databases.
- Prioritizing real-time web results for breaking news or time-sensitive technical evaluations.
- Reducing search latency and costs by utilizing local files or vector caches before hitting external APIs.
- Automating the discovery of counter-evidence to provide balanced, non-biased reports.
- Managing API quotas and fallback logic during high-traffic or unstable network conditions.
How Source Router Works
- Input Analysis: The skill parses the task type, primary subject, and freshness requirements provided by the user or agent.
- Source Prioritization: Based on predefined routing rules, it ranks available sources (web, local_file, vector, graph, counter).
- Bias Mitigation: It checks if counter-evidence is needed based on the goal mode (e.g., comparison or investigation).
- Budget Allocation: Distributes the search budget (maximum rounds) across the selected sources to prevent runaway API costs.
- Stop Rule Generation: Defines specific conditions (e.g., finding 3 authoritative sources) to terminate the search once sufficient data is gathered.
- JSON Plan Output: Produces a structured routing plan including a fallback chain for automated degradation handling.
Source Router Setup
To integrate Source Router into your AI agent environment, ensure the configuration matches the Openclaw Skills standards:
# Example of defining available sources in your environment config
export AVAILABLE_SOURCES=["web", "local_file", "vector", "graph", "counter"]
# Initialize the skill within your agent framework
claw-agent install source-router
Ensure that the fallback chain APIs (Tavily, Bailian, etc.) are correctly configured with valid API keys in your environment variables.
Source Router Data Schema & Taxonomy
Source Router utilizes a strict JSON schema for input and output to ensure interoperability within Openclaw Skills.
Input Specifications
| Field | Required | Description |
|---|---|---|
task_type |
Yes | breaking_news, deep_research, entity_analysis, etc. |
freshness_required |
Yes | high (≤24h), medium (≤30d), low |
available_sources |
Yes | List from [web, local_file, vector, graph, counter] |
Output Structure
task_id: Unique identifier for the routing task.selected_sources: Array of sources with rationale and search hints.fallback_chain: Multi-level degradation logic (e.g., falling back from Tavily to Brave search).degradation_log: A log tracking any failures or source switches during execution.
Source Router Advanced Features
- Dynamic Fallback Chain: Automatically switches between five levels of search providers (Tavily, Bailian, Brave, Direct, Local) based on error triggers like timeouts or quota exhaustion.
- Automated Stop Rules: Context-aware termination logic that stops searching once high-confidence results or specific node counts in a graph are reached.
- Cross-Source Synergy: Ability to identify information gaps in local data and trigger targeted web searches only for the missing pieces.
- Bias Correction: Automatic 'counter' source triggering for comparative tasks to ensure the AI considers opposing viewpoints.
SKILL.md
Loading
Related Openclaw Skills
Medical Translation Skill
A professional-grade translation tool designed for medical terminology and academic writing that prioritizes accuracy and structured execution over general-purpose translation.
Dianping Query
An automated search and data extraction skill for finding restaurant ratings, costs, and locations on Dianping.
PRD to Prototype
An end-to-end automation skill that converts product ideas into professional PRD documents and high-fidelity, interactive HTML/Tailwind prototypes without intermediate questioning.
FBT Travel Assistant
An all-in-one corporate travel assistant for booking hotels and flights with integrated authentication and order management.
Summary Time Keep
A filesystem automation skill that logs local time and generates AI summaries of web content directly into your workspace.
Superpowers Writing Plans
A sophisticated planning skill that transforms high-level requirements into granular, test-driven implementation blueprints for AI agents.
jpeng-helpdesk
A streamlined ticketing system designed to automate support functionality and communication operations.
Featured*
