A technical capability designed to configure, switch, and troubleshoot model providers, routing sessions, and cron executions in OpenClaw.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install hobohan-model-config
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 hobohan-model-config 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 Model Configuration and Routing?
The model-config skill is an essential capability for managing AI model providers, resolving routing conflicts, and handling session model locks within your agent ecosystem. In the realm of Openclaw Skills, ensuring that your agents consistently route to the correct Large Language Model (LLM) is paramount for robust operation, preventing common integration failures like cold-starts or stuck sessions.
By leveraging this skill, developers gain deep architectural control over the model gateway. It streamlines provider setups (such as DeepSeek or MiniMax), handles complex multi-agent routing using dmScope, and provides a structured methodology to switch models dynamically without interrupting existing agent workflows.
Model Configuration and Routing Use Cases
- Setting up new API model providers (such as DeepSeek) directly inside the
openclaw.jsonconfiguration. - Preventing session model lock traps when switching default models on running Telegram agents.
- Pinning critical cron tasks to specific models to avoid execution timeouts caused by cold-start resolutions.
- Routing multi-agent setups (e.g., Patch + Muthu) accurately using scoped Telegram channels.
- Auditing and troubleshooting active model statuses across different sessions and gateway instances.
How Model Configuration and Routing Works
- Provider Resolution: The Gateway scans the config (
openclaw.json) and active plugins, prioritizing plugins over manually configured providers for identical models. - Session Locking: On session creation, a session locks to the active default model (its "birth model"). The session maintains this lock until a manual override occurs or the session expires.
- Cron Pinning: For timed automations, the system overrides default model resolution using explicit model keys within the JSON payload, ensuring deterministic and fast execution.
- Message Routing: Using the
dmScopeconfiguration, messages from multiple bots and users are isolated into distinct sessions to prevent crosstalk.
Model Configuration and Routing Setup
To configure model providers, update your openclaw.json configuration file:
{
"providers": {
"deepseek": {
"apiKey": "sk-...",
"baseUrl": "https://api.deepseek.com"
}
}
}
Verify and check active models or session statuses using the command-line interface:
# Check the current globally active model
openclaw status
# Query the active session-level model
session_status
To list all loaded model plugins and config providers:
openclaw plugin list | grep -i model
grep -A5 '"providers"' ~/.openclaw/openclaw.json
Model Configuration and Routing Data Schema & Taxonomy
The configuration and state are organized across these files and command-line interfaces:
| Configuration Type | Target Location | Description |
|---|---|---|
| Global Gateway Configuration | ~/.openclaw/openclaw.json |
Stores provider keys, base URLs, fallback chains, and routing parameters. |
| Session Status Information | In-memory / Database state | Managed via session_status, tracking individual session-to-model associations. |
| Cron Payload Schema | Scheduled Cron Config | Defines the specific model override (e.g., "model": "deepseek/deepseek-v4-flash") to bypass cold starts. |
Multi-Agent Routing Configuration
Use the following dmScope inside your openclaw.json to organize Telegram multi-agent environments:
{
"telegram": {
"dmScope": "per-channel-peer-account"
}
}
Model Configuration and Routing Advanced Features
- Dynamic Session Switching: Send the
/model <provider/model>command inside an active session to instantly force-switch the underlying model without restarting the gateway. - Multi-Agent Isolation: Configure
dmScopetoper-channel-peer-accountto ensure that parallel bots run on isolated conversation paths. - Time-Sensitive Cron Pinning: Explicitly override model selections inside cron payloads to avoid up to 300-second latency bottlenecks associated with cold-start resolutions of default models.
- Plugin vs. Provider Priority Engine: Automatically resolves model handlers with a hierarchical prioritization strategy (plugin providers > config providers > default fallbacks).
SKILL.md
Loading
Related Openclaw Skills
Agent Browser Shield
A Chromium extension that sanitizes web pages by masking PII, blocking dark patterns, and neutralizing prompt injections before an AI agent parses them.
Expense Tracker Agent Skill
A scheduled, AI-driven daily expense tracker that processes Telegram check-ins, deduplicates entries, and syncs local Markdown ledgers with Google Sheets.
8D Pattern Awareness
This skill transforms unstructured health logs, daily check-ins, or voice transcripts into structured, non-diagnostic wellness pattern maps using the 8 dimensions of wellness.
Cloudflare Developer Platform Skill
A comprehensive reference guide and action plan for building, deploying, and optimizing serverless workflows on the Cloudflare Developer Platform.
MCPorter
MCPorter is a versatile command-line interface designed to configure, authenticate, inspect, and call Model Context Protocol (MCP) servers and tools directly via HTTP or stdio.
Proxmox VE Cluster Management
A powerful skill for AI coding agents to manage Proxmox VE nodes, virtual machines, LXC containers, backups, and storage via the PVE REST API.
Featured*
