openclaw-gpu-bridgeOffload GPU-intensive ML tasks (BERTScore, embeddings) to one or multiple remote GPU machines
Install via ClawdBot CLI:
clawdbot install homeofe/openclaw-gpu-bridgeOpenClaw plugin to offload ML tasks (BERTScore + embeddings) to one or many remote GPU hosts.
hosts[]) with:serviceUrl / url)model / model_type)/status endpoint + batch progress logsgpu_healthgpu_infogpu_status (new in v0.2)gpu_bertscoregpu_embed{
"plugins": {
"@elvatis_com/openclaw-gpu-bridge": {
"hosts": [
{
"name": "rtx-2080ti",
"url": "http://your-gpu-host:8765",
"apiKey": "gpu-key-1"
},
{
"name": "rtx-3090",
"url": "http://your-second-gpu-host:8765",
"apiKey": "gpu-key-2"
}
],
"loadBalancing": "least-busy",
"healthCheckIntervalSeconds": 30,
"timeout": 45,
"models": {
"embed": "all-MiniLM-L6-v2",
"bertscore": "microsoft/deberta-xlarge-mnli"
}
}
}
}{
"plugins": {
"@elvatis_com/openclaw-gpu-bridge": {
"serviceUrl": "http://your-gpu-host:8765",
"apiKey": "gpu-key",
"timeout": 45
}
}
}hosts: array of GPU hosts (v0.2)serviceUrl / url: legacy single-host configloadBalancing: round-robin or least-busyhealthCheckIntervalSeconds: host health polling intervaltimeout: request timeout for compute endpointsapiKey: fallback API key for hosts that do not define per-host keymodels.embed, models.bertscore: plugin-side default modelscd gpu-service
pip install -r requirements.txt
uvicorn gpu_service:app --host 0.0.0.0 --port 8765Default models are warmed on startup:
all-MiniLM-L6-v2microsoft/deberta-xlarge-mnliAdditional models are loaded on-demand and cached in memory.
API_KEY: require X-API-Key for all endpoints except /healthGPU_MAX_CONCURRENT: max parallel jobs (default 2)GPU_EMBED_BATCH: embedding chunk size for progress logging (default 32)MODEL_BERTSCORE: default warm model for BERTScoreMODEL_EMBED: default warm model for embeddingsTORCH_DEVICE: force device (cuda, cpu, cuda:1)GET /healthGET /infoGET /status (queue + active jobs + progress)POST /bertscorePOST /embed/bertscore:
{
"candidates": ["a"],
"references": ["b"],
"model_type": "microsoft/deberta-xlarge-mnli"
}/embed:
{
"texts": ["hello world"],
"model": "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2"
}If you expose your GPU service outside LAN, use defense-in-depth:
API_KEY on serviceapiKey)X-API-Keyserver {
listen 443 ssl http2;
server_name gpu.example.com;
ssl_certificate /etc/letsencrypt/live/gpu.example.com/fullchain.pem;
ssl_certificate_key /etc/letsencrypt/live/gpu.example.com/privkey.pem;
location / {
proxy_pass http://127.0.0.1:8765;
proxy_set_header Host $host;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header X-Forwarded-Proto $scheme;
}
}uvicorn gpu_service:app --host 0.0.0.0 --port 8765 \
--ssl-keyfile /path/key.pem \
--ssl-certfile /path/cert.pemhosts[].urlnpm run build
npm testTypeScript runs in strict mode.
MIT
Generated Mar 1, 2026
A social media or user-generated content platform uses this skill to offload BERTScore calculations for detecting toxic or inappropriate text across multiple posts simultaneously. By leveraging remote GPU hosts, it scales moderation efforts efficiently without overloading local servers, ensuring real-time content safety checks.
An online retail site employs this skill to compute embeddings for product descriptions and user queries, improving search relevance and recommendation systems. The multi-host setup allows handling peak traffic during sales events by distributing GPU-intensive embedding tasks across several machines.
A university research team utilizes this skill to offload BERTScore evaluations for comparing large datasets of scientific papers or student essays. The health checks and failover features ensure reliable processing during long-running analyses, enabling efficient NLP experiments without dedicated local GPU infrastructure.
A customer service company integrates this skill to generate embeddings for support tickets, enabling semantic search and automated response suggestions. The round-robin load balancing distributes tasks across GPU hosts, maintaining low latency for real-time assistance and improving agent productivity.
A legal tech firm uses this skill to compute embeddings for contract clauses and case documents, facilitating similarity searches and compliance checks. The ability to override models per request allows tailoring to specific legal domains, while TLS security ensures sensitive data protection during offloading.
Offer this skill as part of a cloud-based AI service where customers pay a monthly fee based on usage tiers (e.g., number of GPU hosts or compute hours). Revenue is generated from subscriptions that include managed hosting, security features like TLS, and support for multi-GPU scaling.
Sell licenses to enterprises for deploying the skill on their private infrastructure, with pricing based on the number of GPU nodes or concurrent users. Revenue comes from one-time license sales plus optional maintenance contracts for updates and technical support.
Provide custom integration services to help businesses set up and optimize this skill for specific use cases, such as configuring load balancing or securing endpoints. Revenue is generated from project-based fees and ongoing consulting retainers for performance tuning.
💬 Integration Tip
Start with a single GPU host for testing, then scale to multiple hosts using the least-busy load balancing option to optimize performance and reliability in production.
Transform AI agents from task-followers into proactive partners that anticipate needs and continuously improve. Now with WAL Protocol, Working Buffer, Autonomous Crons, and battle-tested patterns. Part of the Hal Stack 🦞
Use the ClawdHub CLI to search, install, update, and publish agent skills from clawdhub.com. Use when you need to fetch new skills on the fly, sync installed skills to latest or a specific version, or publish new/updated skill folders with the npm-installed clawdhub CLI.
Clawdbot documentation expert with decision tree navigation, search scripts, doc fetching, version tracking, and config snippets for all Clawdbot features
Interact with Moltbook social network for AI agents. Post, reply, browse, and analyze engagement. Use when the user wants to engage with Moltbook, check their feed, reply to posts, or track their activity on the agent social network.
OpenClaw CLI wrapper — gateway, channels, models, agents, nodes, browser, memory, security, automation.
MoltGuard — runtime security plugin for OpenClaw agents by OpenGuardrails. Helps users install, register, activate, and check the status of MoltGuard. Use wh...