6wdb

the six-word rule - count them yourself

1Know 2a 3tool 4before 5you 6install.

A database of AI agents, MCP servers, and dev tools - each summed up in a six-word pitch, then spec’d out across the six Ws: who · what · where · when · why · with.

explore the relations map alternatives + works-with, in 3D →

747 tools · page 1 of 32

MCP Server Retrieval & Memory
★ featured
1The 2MCP 3directory 4that 5speaks 6MCP.

6wdb

MCPDirectorySearchRegistry
Application Agents & OrchestrationAutomation & Integration
★ featured
1Open-source 2platform 3for 4building 5LLM 6apps.

Dify

Low-codeLLMOpsRAG
Library / SDK Agents & OrchestrationRetrieval & Memory
★ featured
1The 2original 3toolkit 4for 5LLM 6apps.

LangChain

PythonJavaScriptLLMEcosystem
Library / SDK Agents & Orchestration
★ featured
1Agent 2workflows 3as 4stateful, 5durable 6graphs.

LangGraph

PythonGraphsStateStreaming
Application Agents & OrchestrationCoding
★ featured
1Let 2LLMs 3run 4code 5locally, 6naturally.

Open Interpreter

PythonLocalCode-execution
MCP Server Automation & Integration
★ featured
1Give 2agents 3a 4real, 5driveable 6browser.

Playwright MCP

TypeScriptPlaywrightBrowserDocker
Library / SDK Agents & Orchestration
★ featured
1The 2TypeScript 3toolkit 4for 5AI 6apps.

Vercel AI SDK

TypeScriptReactSDK
Library / SDK Inference & Serving
★ featured
1Run 2LLMs 3efficiently 4on 5any 6hardware.

llama.cpp

C++InferenceLocal
Library / SDK Inference & Serving
★ featured
1Fast, 2high-throughput 3inference 4and 5serving 6engine.

vLLM

PythonInferenceServingGPU
Library / SDK Agents & OrchestrationApps & UIs
1Protocol 2connecting 3AI 4agents 5to 6frontends

AG-UI Protocol

agent-to-ui protocolevent streaminggenerative uiopen standardframework agnostic
Library / SDK Agents & Orchestration
1Open-source 2multi-agent 3framework, 4an 5AutoGen 6fork

AG2

Multi-agentPythonAutoGen
Library / SDK Inference & Serving
1Cloud-native 2infrastructure 3for 4scalable 5LLM 6inference

AIBrix

InferenceKubernetesvLLM
Application Agents & OrchestrationCoding
1AI 2agent 3that 4engineers 5ML 6solutions

AIDE ML

AutoMLAgentPython
Application Apps & UIs
1Browser 2based 3interface 4for 5Stable 6Diffusion

AUTOMATIC1111 Stable Diffusion WebUI

stable-diffusiongradio-uiimage-generationextensionsself-hosted
MCP Server Automation & Integration
1Query 2and 3operate 4AWS 5from 6agents.

AWS MCP Servers

AWSCloudDevOps
Library / SDK Fine-tuning & Training
1Run 2PyTorch 3training 4on 5any 6hardware.

Accelerate

Hugging-FaceDistributedPyTorch
Application Automation & Integration
1Open-source 2workflow 3automation 4with 5AI 6pieces

Activepieces

AutomationWorkflowsLow-code
Library / SDK Agents & Orchestration
1Build 2and 3auto-optimize 4LLM 5task 6pipelines

AdalFlow

OptimizationPipelinesPython
Hosted Service Observability & EvaluationGuardrails & Structure
1Turns 2production 3AI 4traces 5into 6evals

Adaline

llm-opsevaluationprompt-managementproduction-tracessynthetic-data
Library / SDK Agents & Orchestration
1Multi 2agent 3agency 4orchestration 5in 6Python

Agency Swarm

Multi-agentPythonOpenAIOrchestration
Application Agents & OrchestrationAutomation & Integration
1An 2agent 3that 4operates 5your 6computer.

Agent S

GUI-agentComputer-useAutonomous
Library / SDK Agents & Orchestration
1Routes 2conversations 3across 4multiple 5specialized 6agents

Agent Squad

multi-agentintent-routingawsconversation-stateorchestration
Application Agents & OrchestrationAutomation & Integration
1A 2personal, 3general-purpose 4autonomous 5agent 6framework

Agent Zero

AutonomousGeneral-purposePython
Application Agents & OrchestrationAutomation & Integration
1Configure 2and 3deploy 4autonomous 5agents 6online.

AgentGPT

TypeScriptBrowserAutonomous

not a gimmick, a context budget

Six words and six Ws is the smallest honest unit an agent needs to pick the right tool.

Every entry is deliberately token-lean, so an LLM can shortlist and commit to the correct MCP server or tool mid-task without pulling a README into its context window. Reads are open JSON and there is a native MCP endpoint, so agents can query and compare mid-loop, then read a Watch note on install-risk before committing.

  • → Token-lean entries, not thousand-word READMEs
  • → Query mid-loop over MCP or the open JSON API
  • → Each entry carries a Watch note on install-risk

for agents & scripts

Reading this as a machine? Query it directly.

Search is open JSON - no key. Report telemetry after using a tool and it feeds that tool’s Proof Score. Or speak MCP to /mcp and discover tools mid-loop.