convex-backend

What This Skill Does

The convex-backend skill provides a robust, cloud-native storage architecture for your OpenClaw agent. By replacing ephemeral local files with a persistent Convex backend, you ensure that your agent's memory, secrets, and daily logs remain intact across container restarts, environment wipes, and multi-device deployments. This skill leverages the official Convex MCP server to provide a standardized, secure API for state management, effectively bridging the gap between local execution environments and centralized cloud storage.

Installation

1. Install the skill: clawhub install openclaw/skills/skills/launchthatbot/launchthatbot-convex-backend
2. Navigate to the installation directory: cd /home/node/.openclaw/skills/convex-backend
3. Deploy the backend components: CONVEX_DEPLOY_KEY=your_key_here npx -y convex@latest deploy
4. Configure your MCP settings by adding the convex server block to your agent configuration file, pointing it to the convex@latest mcp start command.
5. Verify compatibility using mcporter list convex --schema. If mcporter is not found, ensure it is installed globally or via npx.

Use Cases

• Long-term Agent Memory: Persist task progress and historical context over days or weeks of operation without fear of local storage loss.
• Centralized Credential Management: Securely store API keys and service tokens in Convex environment variables, accessible to the agent across different runtime environments.
• Audit Logging: Automatically save session summaries and daily logs to a remote database, enabling external review via the LaunchThatBot dashboard.
• Multi-Instance Sync: Synchronize state across multiple agent instances by pointing them toward the same Convex deployment.

Example Prompts

1. "Save my current progress on the API integration task to the daily logs in Convex."
2. "Store my OpenAI API key in the secure environment variables so I don't have to provide it every session."
3. "Check if there are any existing credentials for the GitHub service in Convex before starting the deployment process."

Tips & Limitations

• CRITICAL: Never store the CONVEX_DEPLOY_KEY inside the Convex backend itself; it must remain local.
• Security: Always prefer envSet for credentials over saving them into plain text files like MEMORY.md.
• Dependencies: Ensure your network allows connections to Convex endpoints, as the agent requires real-time cloud access for all read/write operations.
• Bridging: If using restricted runtimes, remember to utilize mcporter to execute Convex tools correctly.