OpenClaw VSR Bridge

When To Use

• Use this skill when an OpenClaw agent should install VSR without auto-launching the dashboard.
• Use this skill when an existing openclaw.json should be moved under VSR management.
• Use this skill when you want the OpenClaw model base URL rewritten only after canonical config.yaml is written successfully.

Requirements

• curl
• bash
• write access to the OpenClaw config path and the target VSR config path

Workflow

1. Check whether VSR is already installed with command -v vllm-sr.

2. If VSR is missing, run the supported installer in agent-safe mode:

   curl -fsSL https://vllm-sr.ai/install.sh | bash -s -- --mode cli --runtime skip --no-launch

3. Locate the OpenClaw config in this order unless the user already gave you a path:

   • $OPENCLAW_CONFIG_PATH
   • ./openclaw.json
   • ~/.openclaw/openclaw.json

4. Import the OpenClaw model providers into canonical VSR config:

   vllm-sr config import --from openclaw --source <openclaw.json> --target config.yaml

5. Do not pass --force unless the user explicitly approves overwriting existing backup files.

6. After a successful import, report:

   • the imported logical model names
   • the canonical target config path
   • the rewritten OpenClaw config path
   • the rewritten OpenClaw base URL
   • the backup file paths, if any

7. Suggest the next validation commands:

   vllm-sr config validate --config config.yaml
   vllm-sr config router --config config.yaml

8. If the user wants runtime verification after validation, suggest:

   vllm-sr serve --config config.yaml
   curl http://127.0.0.1:8899/v1/models

Boundaries

• Do not invent a second installer path. Use the supported installer above.
• Do not rewrite OpenClaw before the VSR target config has been written successfully.
• Do not import OpenClaw rooms, teams, browser settings, command settings, or memory history into VSR.