CCM
/MCP
SkillsMCPMarketplacesDigestToolsAdvertise

This week in Claude

Every Monday: Claude Code, Agent SDK, MCP, and the Anthropic platform moves worth your time.

Skills by Category
Frontend DevelopmentBackend & APIsTesting & QASecurityDevOps & CI/CDGit & Pull RequestsDocumentationCode Review & QualityAI & Agent BuildingSkill Development
MCP Servers by Category
Sales & MarketingWeb & Browser AutomationDatabasesAI & LLM ToolsCloud & InfrastructureCommunication & MessagingDeveloper ToolsDesign & CreativeDocuments & KnowledgeSearch & Web Crawling
Marketplaces by Category
AI Agents & OrchestrationLLM IntegrationDevelopment ToolsFrontend & UIBackend & APIsDatabasesTesting & Code QualityDevOps & CloudSecurity & ComplianceGit & Version Control

Claude Code Marketplaces

Discover Claude Code plugins, extensions, and tools. Automatically updated directory of Anthropic Claude AI marketplaces with development tools, productivity plugins, and integrations.

Resources

  • Browse Skills
  • Browse MCP Servers
  • Browse Marketplaces
  • Skill index
  • MCP index
  • Marketplace index
  • Plugins Reference

Community

  • About
  • Tools
  • Feedback
  • Privacy Policy
  • Advertise

Built for the Claude Code community with Claude Code by mertbuilds.com

Independent project, not affiliated with Anthropic
buggy1111 avatar

Shelly Smart Home

buggy1111/shelly-mcp
STDIOregistry active
Summary

Local-first control for Shelly smart home devices across all generations (Gen1 through Gen4). Connects directly to devices on your LAN with ~10ms latency and full API access, plus optional cloud fallback for remote use. Beyond basic switch and light control, you get energy monitoring with history, schedule management, on-device script execution, webhooks, and a key-value store for automation. The generic RPC engine exposes every device capability, even for hardware released after the server shipped. Scene definitions let you save multi-device routines as named commands. Mutations are audit-logged and destructive operations require explicit confirmation gates to prevent accidental factory resets or data loss.

CodeRabbit
CodeRabbit
AI writes the code. CodeRabbit catches the slop.
Try For Free →
ego lite browserego lite browser
ego lite browser
Fastest browser for AI agents to run web automation tasks, always free.
Download Free life-time →
CodeHealth MCP ServerCodeHealth MCP Server
CodeHealth MCP Server
Protect your code quality, stop the AI slop.
Try For Free →
belt - the only tool your agent needs
belt - the only tool your agent needs
belt cli automatically finds the best tools and skills for your agent. image, video, music, tts...
one prompt install →
Open Steps
Open Steps
Free an open-source skills that make AI coding agents easier to understand, verify, and control.
Download for free →
AppSignal
AppSignal
Monitor with ease. Code with confidence.
Start Free Trial →
Agent, connect blockchain
Agent, connect blockchain
Connect your Claude agent to live crypto prices and trading routes via 1inch
Get the MCP →
Granola, the best AI meeting recorder
Granola, the best AI meeting recorder
Notes, actions and memory. Without a meeting bot. First month 100% off.
Download for free →
CodeRabbit
CodeRabbit
AI writes the code. CodeRabbit catches the slop.
Try For Free →
ego lite browserego lite browser
ego lite browser
Fastest browser for AI agents to run web automation tasks, always free.
Download Free life-time →
CodeHealth MCP ServerCodeHealth MCP Server
CodeHealth MCP Server
Protect your code quality, stop the AI slop.
Try For Free →
belt - the only tool your agent needs
belt - the only tool your agent needs
belt cli automatically finds the best tools and skills for your agent. image, video, music, tts...
one prompt install →
Open Steps
Open Steps
Free an open-source skills that make AI coding agents easier to understand, verify, and control.
Download for free →
AppSignal
AppSignal
Monitor with ease. Code with confidence.
Start Free Trial →
Agent, connect blockchain
Agent, connect blockchain
Connect your Claude agent to live crypto prices and trading routes via 1inch
Get the MCP →
Granola, the best AI meeting recorder
Granola, the best AI meeting recorder
Notes, actions and memory. Without a meeting bot. First month 100% off.
Download for free →

shelly-mcp

mcp-name: io.github.Buggy1111/shelly-mcp

MCP server for the entire Shelly smart-home ecosystem — read, control, and automate Shelly devices of every generation (Gen1 → Gen4; BLU via its gateway or the generic RPC engine, dedicated BLU tools are on the roadmap) from any MCP client (Claude Desktop, Claude Code, Cursor, …). Local-first (zero rate-limit, ~10 ms, full API), with cloud fallback for off-LAN access.

⚠️ Unofficial community project. Not affiliated with, endorsed by, or sponsored by Allterco Robotics / Shelly. "Shelly" is a trademark of its respective owner.

🚧 Alpha / in development. See docs/ for the full design.

Why

The only existing Shelly MCP servers are cloud-only and minimal. shelly-mcp unifies Gen1 and Gen2+ behind one tool surface, covers energy monitoring and automation (schedules, scripts, webhooks, KVS), and auto-discovers any device's capabilities — including hardware released after this server was written.

Install

uvx shelly-mcp          # or: pip install shelly-mcp

Register in your MCP client:

{ "mcpServers": { "shelly": { "command": "uvx", "args": ["shelly-mcp"] } } }

Configure

Auto-discovery (mDNS) finds devices on your local subnet — shelly_discover and you're running. For named devices ("turn off the kitchen"), devices on other subnets, passwords, or the cloud fallback, create ~/.config/shelly-mcp/config.yaml (chmod 600):

devices:
  televize:
    ip: 192.168.0.101
    location: obývák            # lets "turn off the living room" work
    # password: "..."           # only if the device has auth (or env SHELLY_PW_televize)
cloud:
  enabled: false                # optional off-LAN fallback
  # auth_key via env SHELLY_CLOUD_AUTH_KEY

Full reference: config.example.yaml and docs/04-CONFIG-AND-DEPLOY.md.

What data leaves your machine

None, beyond the calls to your own Shelly devices (on your LAN) and — only if you enable it — your own Shelly Cloud account. No telemetry, no phone-home.

Capabilities (local vs cloud)

LocalCloud
Discovery, status, control✅⚠️ control + status only
Energy live + history✅⚠️ live only
Automation (schedules/scripts/webhooks/KVS)✅❌

Tools, resources & prompts

Read (safe): shelly_version · shelly_discover · shelly_list_devices · shelly_get_info · shelly_get_status (normalized) · shelly_get_config (credentials masked) · shelly_list_components · shelly_list_methods

Control (audited): shelly_switch_set · shelly_switch_toggle · shelly_light_set (RGBW/CCT/white) · shelly_cover_move

Energy: shelly_energy_live · shelly_energy_history

Generic engine (total coverage): shelly_rpc (read-only) · shelly_rpc_write (mutations, confirm:true + data-loss double-gate)

System / schedules (gated): shelly_system_reboot|update|set_auth · shelly_schedule_list|create|update|delete

Automation (Gen2+ local-only): shelly_kvs_* (key-value store) · shelly_webhook_* (event→HTTP) · shelly_script_* (on-device JS — list/get_code/create/put_code/start/stop/eval/delete, chunked upload) · shelly_virtual_* (virtual components). Deletes + arbitrary-code paths (script_put_code/eval) are confirm:true-gated.

Scenes (deterministic, named): shelly_scene_list|get|run|create|delete — define a multi-device routine once and run it by name (shelly_scene_run "film"), identical every time and schedulable from any client. Stored in ~/.config/shelly-mcp/scenes.yaml (see scenes.example.yaml); scenes and schedules accept only plain control methods (Switch/Light/RGB(W)/CCT/Cover) — never Script.Eval, SetAuth, or anything destructive (ADR-007, docs/06-SCENES.md).

Resources: shelly://devices, shelly://device/{name}/status — Prompts: shelly_evening_scene, shelly_energy_report, shelly_diagnose

Safety: reads are readOnlyHint; every mutation is audit-logged; the generic write tool and destructive system tools require explicit confirm:true, and irreversible methods (factory reset, wipe-all) need a second i_understand_data_loss gate — so even a hijacked LLM can't silently destroy a device.

Troubleshooting

  • "Device unreachable" — confirm the IP (shelly_discover, your router's client list, or the Shelly app → device → Settings → Device information), and that the machine running the server is on the same LAN. In WSL/containers, mDNS discovery usually doesn't work — configure devices by ip in the config file instead (that path needs no mDNS).
  • "Auth required" — the device has a password: add password: under the device in the config (or SHELLY_PW_<name> env var).
  • Cloud-only device refuses automation tools — expected: the Shelly Cloud API can't manage schedules/scripts/webhooks/KVS; connect locally for those.
  • Config refuses to load — if it contains a secret, it must be chmod 600 (deliberate, fail-closed).

Docs

Full design in docs/ (indexed): overview, architecture (+ADRs), tool surface, security, scenes, config/deploy, build plan, project log, roadmap, the launch runbook, and the complete Shelly API catalog.

License

MIT — see LICENSE.

Featured
CodeRabbit
CodeRabbit
AI writes the code. CodeRabbit catches the slop.
Try For Free →
ego lite browserego lite browser
ego lite browser
Fastest browser for AI agents to run web automation tasks, always free.
Download Free life-time →
CodeHealth MCP ServerCodeHealth MCP Server
CodeHealth MCP Server
Protect your code quality, stop the AI slop.
Try For Free →
belt - the only tool your agent needs
belt - the only tool your agent needs
belt cli automatically finds the best tools and skills for your agent. image, video, music, tts...
one prompt install →
Open Steps
Open Steps
Free an open-source skills that make AI coding agents easier to understand, verify, and control.
Download for free →
AppSignal
AppSignal
Monitor with ease. Code with confidence.
Start Free Trial →
Agent, connect blockchain
Agent, connect blockchain
Connect your Claude agent to live crypto prices and trading routes via 1inch
Get the MCP →
Granola, the best AI meeting recorder
Granola, the best AI meeting recorder
Notes, actions and memory. Without a meeting bot. First month 100% off.
Download for free →
Categories
Automation & Workflows
Registryactive
Packageshelly-mcp
TransportSTDIO
UpdatedJun 10, 2026
View on GitHub

More from buggy1111

  • Czech PII Anonymizer & NLP

Related Automation & Workflows MCP Servers

View all →
build.uplink avatar
Uplink

build.uplink/uplink

Automate a real, logged-in browser on your own device — no stored credentials, no bot detection.
ceki-me avatar
Mcp Server

ceki-me/mcp-server

Hire specialists by the hour — search, schedule, and pay via MCP protocol.
ciprianpater avatar
srv-d7aoqmh5pdvs7391dcqg

ciprianpater/srv-d7aoqmh5pdvs7391dcqg

# NWO Robotics MCP Server Control real robots, IoT devices, and autonomous agent swarms through natural language — powered by the [NWO Robotics API](https://nwo.capital). --- ## What This Server Does This MCP server exposes the full NWO Robotics API as 64 ready-to-use tools. Any MCP-compatible AI agent (Claude, ChatGPT, Cursor, etc.) can use it to: - Send natural language instructions to physical robots - Run Visual-Language-Action (VLA) inference on live camera feeds - Plan, validate, and execute multi-step robot tasks - Monitor sensors, detect slip, and fuse multi-modal data - Train robots online with reinforcement learning - Register and manage agent identities on Base mainnet via the Cardiac biometric ID system No local installation needed. The server runs on Render and is ready to connect. --- ## Tools Overview ### 🤖 VLA Inference & Models Run Vision-Language-Action inference on any supported robot. Send a text instruction and camera images, receive joint action vectors in real time. Supports auto model routing, ultra-low-latency Cloudflare edge inference (28ms avg), and WebSocket streaming at up to 50Hz. `vla_inference` · `edge_inference` · `list_models` · `get_model_info` · `get_streaming_config` --- ### 🦾 Robot Control & State Query live robot state (joint angles, gripper, battery, position), execute pre-computed action sequences, and fuse camera + lidar + thermal + force + GPS sensor inputs into a single inference call. `query_robot_state` · `execute_actions` · `sensor_fusion` · `robot_query` · `get_agent_status` --- ### 🗺️ Task Planning & Learning Decompose complex instructions into ordered subtasks, execute them step by step, poll progress, and log outcomes so the model learns and improves with every run. `task_planner` · `execute_subtask` · `status_poll` · `learning_recommend` · `learning_log` --- ### 🔑 Agent Management Self-register a new AI agent in under 2 seconds, check your monthly API quota, upgrade tiers by paying ETH, and manage robot registrations and capabilities. | Tier | Calls/month | Cost | |------|-------------|------| | Free | 100,000 | $0 | | Prototype | 500,000 | ~0.015 ETH/mo | | Production | Unlimited | ~0.062 ETH/mo | `register_agent` · `check_balance` · `pay_upgrade` · `create_wallet` · `register_robot` · `update_agent` · `get_agent_info` --- ### 🔍 Agent Discovery Discover all available execution modes (mock / simulated / live), robot types, VLA models, and sensor capabilities. Validate tasks with a dry-run before committing to execution. `nwo_health` · `nwo_whoami` · `discover_capabilities` · `dry_run` · `plan_task` --- ### 🔌 ROS2 Bridge (Physical Robots) Connect directly to physical robots over the ROS2 bridge. Send joint commands, submit action sequences, and trigger emergency stops on one or all robots within 10ms. Supported: UR5e, Panda, Spot, Unitree G1, and more. `ros2_list_robots` · `ros2_robot_status` · `ros2_send_command` · `ros2_submit_action` · `ros2_emergency_stop` · `ros2_emergency_stop_all` · `ros2_get_robot_types` --- ### 🧪 Physics Simulation Simulate trajectories, check for collisions, estimate joint torques, validate grasps, and plan collision-free motions with MoveIt2 — before touching real hardware. `simulate_trajectory` · `check_collision` · `estimate_torques` · `validate_grasp` · `plan_motion` · `get_scene_library` · `generate_scene` --- ### 📐 Embodiment & Calibration Browse the robot embodiment registry (DOF, joint limits, sensors), download URDF models, get normalization parameters for VLA inference, and run automatic joint calibration. `list_embodiments` · `get_robot_specs` · `get_normalization` · `download_urdf` · `get_test_results` · `compare_robots` · `run_calibration` · `calibrate_confidence` --- ### 🧠 Online RL & Fine-Tuning Start online reinforcement learning sessions, stream state/action/reward telemetry, build fine-tuning datasets from logged runs, and launch LoRA fine-tuning jobs on any base VLA model. `start_rl_training` · `submit_rl_telemetry` · `create_finetune_dataset` · `start_finetune_job` --- ### 🖐️ Tactile Sensing (ORCA Hand) Read 256-taxel tactile sensor arrays from the ORCA robot hand, assess grip quality and object texture, and detect slip in real time to prevent dropped objects. `read_tactile` · `process_tactile` · `detect_slip` --- ### 📦 Dataset Hub Access 1.54 million+ human robot demonstrations for the Unitree G1 humanoid (430+ hours, LeRobot-compatible format) for training and fine-tuning. `list_datasets` --- ### 🫀 Cardiac Blockchain Identity (Base Mainnet) Register AI agents on Base mainnet and receive a permanent soul-bound Digital ID (`rootTokenId`). Issue verifiable credentials for task authorization, swarm control, location access, and payments — all gasless via the NWO relayer. Smart contracts deployed on Base Mainnet (Chain ID 8453): - `NWOIdentityRegistry` — `0x78455AFd5E5088F8B5fecA0523291A75De1dAfF8` - `NWOAccessController` — `0x29d177bedaef29304eacdc63b2d0285c459a0f50` - `NWOPaymentProcessor` — `0x4afa4618bb992a073dbcfbddd6d1aebc3d5abd7c` `cardiac_register_agent` · `cardiac_identify_agent` · `cardiac_renew_key` · `cardiac_issue_credential` · `cardiac_check_credential` · `cardiac_grant_access` · `cardiac_get_nonce` · `cardiac_check_access` · `cardiac_payment_process` --- ### 🔮 Cardiac Oracle Validate ECG biometric data from smartwatches to authenticate human identities, compute cardiac hashes, and verify recent validations. `oracle_health` · `oracle_validate_ecg` · `oracle_hash_ecg` · `oracle_verify` --- ## Supported Robot Models | Model | Type | Capabilities | |-------|------|--------------| | `xiaomi-robotics-0` | VLA | Grasp, navigate, manipulate | | `pi05` | VLA | General manipulation | | `groot_n1.7` | VLA | Humanoid control | | `deepseek-ocr-2b` | OCR | Label reading, text recognition | --- ## Example Usage **Pick and place:** > "Pick up the red box from the table and place it on shelf B" **Sensor query:** > "What is the temperature in warehouse zone 3?" **Safety:** > "Run a safety check before moving robot_001 to the loading dock" **Swarm:** > "Deploy all available robots to patrol the perimeter" **Learning:** > "What grip technique should I use for fragile glass objects?" --- ## Links - 🌐 [NWO Capital](https://nwo.capital) - 📄 [Agent Skill File](https://nwo.capital/webapp/agent.md) - 📖 [API Docs](https://nwo.capital/webapp/nwo-robotics.html) - 🧬 [Cardiac SDK](https://github.com/RedCiprianPater/nwo-cardiac-sdk) - 🔑 [Get API Key](https://nwo.capital/webapp/api-key.php) - 🤗 [Live Demo](https://huggingface.co/spaces/PUBLICAE/nwo-robotics-api-demo) - 📜 [OpenAPI Spec](https://nwo.capital/openapi.yaml) --- ## Support 📧 support@nwo.capital
clauxel avatar
Agenticbudgetrouter Mcp

clauxel/agentic-budget-router-mcp

Agentic workflow budget approvals with usage receipts.
co.pipeboard avatar
Google Ads Mcp

co.pipeboard/google-ads-mcp

Google Ads automation with AI: analyze performance, manage campaigns, optimize bids.
co.pipeboard avatar
Meta Ads Mcp

co.pipeboard/meta-ads-mcp

Facebook / Meta Ads automation with AI: analyze performance, test creatives, optimize spend.