Index live · v1.7.0 · SEP 12 2026

192+ skills indexed200+ MCP servers9 platforms tracked

04 · Developer Tools

The stack
underneath
the agent.

Skills and MCP servers describe what an agent can do. These are the frameworks that orchestrate it, the sandboxes that run its code, and the tooling that makes it observable in production.

Browse Tools How We Select

the layersstack

# what the agent knows
skills      → /skills

# what it can reach
mcp         → /mcp

# what runs it
framework   → orchestration
sandbox     → isolation
tracing     → observability
memory      → persistence

The index

Every tool we index

55 of 55

LangGraph

by LangChain

Graph-based orchestration with durable, inspectable state for long-running agents.

HybridPythonTypeScript

Agent Frameworks

Open sourceOpen

Pydantic AI

by Pydantic

Type-safe agents with validated structured output and dependency injection.

Self-hostedPython

Agent Frameworks

Open sourceOpen

CrewAI

by CrewAI

Role-based multi-agent orchestration — agents as a crew with assigned responsibilities.

Agent Frameworks

Open sourceOpen

Mastra

by Mastra

TypeScript-native agent framework built for Next.js and edge deployment.

HybridTypeScript

Agent Frameworks

Open sourceOpen

LlamaIndex

by LlamaIndex

Event-driven workflows and the data layer for retrieval-heavy agents.

HybridPythonTypeScript

Agent Frameworks

Open sourceOpen

smolagents

by Hugging Face

Minimal agents that act by writing Python, from Hugging Face.

Self-hostedPython

Agent Frameworks

Open sourceOpen

OpenAI Agents SDK

by OpenAI

Lightweight agent loop with handoffs, guardrails, and tracing built in.

Self-hostedPythonTypeScript

Agent Frameworks

Open sourceOpen

E2B

by E2B

Firecracker microVM sandboxes purpose-built for running AI-generated code.

HybridPythonTypeScript

Modal

by Modal

Serverless Python compute with GPUs, built for inference and heavy ML workloads.

Daytona

by Daytona

Fast, stateful development environments for agents that need to persist a workspace.

HybridGoTypeScript

Open sourceOpen

Northflank

by Northflank

Production platform with microVM isolation and bring-your-own-cloud deployment.

Usage-basedOpen

Fly.io

by Fly.io

Firecracker VMs at the edge, with no ephemeral time limit on long-running agents.

Usage-basedOpen

Cloudflare Workers

by Cloudflare

Edge compute with container support, Durable Objects, and R2 storage.

HostedTypeScriptJavaScript

Claude Code

by Anthropic

Anthropic's agentic CLI — reads a codebase, edits across files, and runs the tests.

Coding Agents & CLIs

SubscriptionOpen

Gemini CLI

by Google

Google's open-source terminal agent, with a generous free tier.

Self-hostedTypeScript

Coding Agents & CLIs

Codex CLI

by OpenAI

OpenAI's open-source coding agent for the terminal.

Self-hostedRustTypeScript

Coding Agents & CLIs

SubscriptionOpen

Aider

by Aider

Model-agnostic pair programming in the terminal, with git-native commits.

Self-hostedPython

Coding Agents & CLIs

Open sourceOpen

Cline

by Cline

Autonomous coding agent inside VS Code, with MCP support and human approval gates.

Self-hostedTypeScript

Coding Agents & CLIs

Open sourceOpen

Cursor

by Anysphere

AI-native code editor built on VS Code — agent mode, inline edits, and MCP servers in the IDE.

Coding Agents & CLIs

Windsurf

by Windsurf

Agentic IDE with Cascade — an agent that tracks what you did across the session, not just the current prompt.

Coding Agents & CLIs

Official MCP Registry

by Model Context Protocol

The community registry maintained by the Model Context Protocol project.

MCP Infrastructure

Open sourceOpen

Smithery

by Smithery

MCP registry and hosting platform, with local install or managed endpoints.

HybridTypeScript

MCP Infrastructure

Glama

by Glama

MCP directory with automated server inspection and quality grading.

HostedTypeScript

MCP Infrastructure

Docker MCP Catalog

by Docker

Containerised MCP servers with signed images and a managed gateway.

MCP Infrastructure

Langfuse

by Langfuse

Open-source LLM tracing, evaluation, and prompt management.

HybridPythonTypeScript

Observability & Evals

Open sourceOpen

LangSmith

by LangChain

Tracing, evaluation, and monitoring from the LangChain team.

HybridPythonTypeScript

Observability & Evals

Phoenix

by Arize AI

Open-source, OpenTelemetry-native tracing and evaluation for LLM apps.

HybridPythonTypeScript

Observability & Evals

Open sourceOpen

Braintrust

by Braintrust

Evaluation-first platform for shipping LLM changes with measured confidence.

HybridPythonTypeScript

Observability & Evals

Helicone

by Helicone

Open-source LLM gateway with logging, caching, and cost tracking.

HybridTypeScript

Observability & Evals

Open sourceOpen

Mem0

by Mem0

A memory layer that extracts and recalls durable facts across sessions.

HybridPythonTypeScript

Open sourceOpen

Zep

by Zep

Temporal knowledge graph memory that tracks how facts change over time.

HybridPythonTypeScript

Open sourceOpen

Letta

by Letta

Stateful agent runtime with self-editing memory, from the MemGPT research.

HybridPythonTypeScript

Open sourceOpen

Supabase

by Supabase

Postgres with auth, storage, realtime, and edge functions in one open-source platform.

HybridTypeScriptPython

Open sourceOpen

Neon

by Neon

Serverless Postgres with database branching and scale-to-zero.

HostedTypeScriptPython

PlanetScale

by PlanetScale

MySQL platform built on Vitess, with branching and non-blocking schema changes.

HostedTypeScriptPython

Usage-basedOpen

Convex

by Convex

FSL-1.1-Apache-2.0

Reactive backend where queries, mutations, and scheduling are TypeScript functions.

HybridTypeScript

Turso

by Turso

SQLite-compatible libSQL at the edge, cheap enough for a database per tenant.

HybridTypeScriptPython

Upstash

by Upstash

Serverless Redis, Kafka, and vector search billed per request.

HostedTypeScriptPython

MongoDB

by MongoDB

Document database with managed Atlas hosting and native vector search.

HybridTypeScriptPython

ClickHouse

by ClickHouse

Columnar OLAP database for analytics over very large event tables.

HybridSQLPython

Open sourceOpen

Firebase

by Google

Google's app backend — realtime data, auth, functions, and hosting.

HostedTypeScriptPython

Pinecone

by Pinecone

Fully managed vector database with no index infrastructure to operate.

HostedPythonTypeScript

Vector & Retrieval

Qdrant

by Qdrant

Open-source vector database in Rust, with rich payload filtering.

HybridPythonTypeScript

Vector & Retrieval

Open sourceOpen

Weaviate

by Weaviate

Open-source vector database with built-in hybrid search and vectorizer modules.

HybridPythonTypeScript

Vector & Retrieval

Open sourceOpen

Chroma

by Chroma

Embeddable open-source vector store that runs in-process for local development.

HybridPythonTypeScript

Vector & Retrieval

Open sourceOpen

Milvus

by Zilliz

Distributed open-source vector database built for billion-scale collections.

HybridPythonTypeScript

Vector & Retrieval

Open sourceOpen

n8n

by n8n

Sustainable Use License (source-available)

Source-available workflow automation with native AI agent nodes, MCP support, and a self-host option.

HybridTypeScriptJavaScript

Orchestration & Workflows

Temporal

by Temporal Technologies

Durable execution — workflows that survive process crashes, deploys, and week-long waits for a human.

Orchestration & Workflows

Open sourceOpen

Trigger.dev

by Trigger.dev

Open-source background jobs for TypeScript — long-running agent tasks without timeouts, with retries and realtime status.

HybridTypeScript

Orchestration & Workflows

Open sourceOpen

Vercel

by Vercel

Frontend cloud for Next.js and the AI SDK — serverless and streaming functions with an official MCP server.

HostedTypeScriptJavaScript

Hosting & Deployment

RunPod

by RunPod

GPU cloud for inference and fine-tuning — on-demand pods and serverless endpoints that scale to zero.

HostedPythonAny

Hosting & Deployment

Usage-basedOpen

DigitalOcean

by DigitalOcean

General-purpose cloud with plain pricing — Droplets, managed databases, a PaaS, and GPU instances.

Hosting & Deployment

Usage-basedOpen

Auth for GenAI

by Auth0 / Okta

Token vaulting and user-approved delegated access for agents calling APIs.

HostedPythonTypeScript

Auth & Governance

Oso

by Oso

Authorization as a service, with a policy language for fine-grained access.

HybridPythonTypeScript

Auth & Governance

Permit.io

by Permit.io

Policy-as-code authorization with an access-request layer for agent actions.

HybridPythonTypeScript

Auth & Governance

Complete Developer Tool Index

LangGraph — Graph-based orchestration with durable, inspectable state for long-running agents.
Pydantic AI — Type-safe agents with validated structured output and dependency injection.
CrewAI — Role-based multi-agent orchestration — agents as a crew with assigned responsibilities.
Mastra — TypeScript-native agent framework built for Next.js and edge deployment.
LlamaIndex — Event-driven workflows and the data layer for retrieval-heavy agents.
smolagents — Minimal agents that act by writing Python, from Hugging Face.
OpenAI Agents SDK — Lightweight agent loop with handoffs, guardrails, and tracing built in.
E2B — Firecracker microVM sandboxes purpose-built for running AI-generated code.
Modal — Serverless Python compute with GPUs, built for inference and heavy ML workloads.
Daytona — Fast, stateful development environments for agents that need to persist a workspace.
Northflank — Production platform with microVM isolation and bring-your-own-cloud deployment.
Fly.io — Firecracker VMs at the edge, with no ephemeral time limit on long-running agents.
Cloudflare Workers — Edge compute with container support, Durable Objects, and R2 storage.
Claude Code — Anthropic's agentic CLI — reads a codebase, edits across files, and runs the tests.
Gemini CLI — Google's open-source terminal agent, with a generous free tier.
Codex CLI — OpenAI's open-source coding agent for the terminal.
Aider — Model-agnostic pair programming in the terminal, with git-native commits.
Cline — Autonomous coding agent inside VS Code, with MCP support and human approval gates.
Cursor — AI-native code editor built on VS Code — agent mode, inline edits, and MCP servers in the IDE.
Windsurf — Agentic IDE with Cascade — an agent that tracks what you did across the session, not just the current prompt.
Official MCP Registry — The community registry maintained by the Model Context Protocol project.
Smithery — MCP registry and hosting platform, with local install or managed endpoints.
Glama — MCP directory with automated server inspection and quality grading.
Docker MCP Catalog — Containerised MCP servers with signed images and a managed gateway.
Langfuse — Open-source LLM tracing, evaluation, and prompt management.
LangSmith — Tracing, evaluation, and monitoring from the LangChain team.
Phoenix — Open-source, OpenTelemetry-native tracing and evaluation for LLM apps.
Braintrust — Evaluation-first platform for shipping LLM changes with measured confidence.
Helicone — Open-source LLM gateway with logging, caching, and cost tracking.
Mem0 — A memory layer that extracts and recalls durable facts across sessions.
Zep — Temporal knowledge graph memory that tracks how facts change over time.
Letta — Stateful agent runtime with self-editing memory, from the MemGPT research.
Supabase — Postgres with auth, storage, realtime, and edge functions in one open-source platform.
Neon — Serverless Postgres with database branching and scale-to-zero.
PlanetScale — MySQL platform built on Vitess, with branching and non-blocking schema changes.
Convex — Reactive backend where queries, mutations, and scheduling are TypeScript functions.
Turso — SQLite-compatible libSQL at the edge, cheap enough for a database per tenant.
Upstash — Serverless Redis, Kafka, and vector search billed per request.
MongoDB — Document database with managed Atlas hosting and native vector search.
ClickHouse — Columnar OLAP database for analytics over very large event tables.
Firebase — Google's app backend — realtime data, auth, functions, and hosting.
Pinecone — Fully managed vector database with no index infrastructure to operate.
Qdrant — Open-source vector database in Rust, with rich payload filtering.
Weaviate — Open-source vector database with built-in hybrid search and vectorizer modules.
Chroma — Embeddable open-source vector store that runs in-process for local development.
Milvus — Distributed open-source vector database built for billion-scale collections.
n8n — Source-available workflow automation with native AI agent nodes, MCP support, and a self-host option.
Temporal — Durable execution — workflows that survive process crashes, deploys, and week-long waits for a human.
Trigger.dev — Open-source background jobs for TypeScript — long-running agent tasks without timeouts, with retries and realtime status.
Vercel — Frontend cloud for Next.js and the AI SDK — serverless and streaming functions with an official MCP server.
RunPod — GPU cloud for inference and fine-tuning — on-demand pods and serverless endpoints that scale to zero.
DigitalOcean — General-purpose cloud with plain pricing — Droplets, managed databases, a PaaS, and GPU instances.
Auth for GenAI — Token vaulting and user-approved delegated access for agents calling APIs.
Oso — Authorization as a service, with a policy language for fine-grained access.
Permit.io — Policy-as-code authorization with an access-request layer for agent actions.

By layer

Browse by Category

11 layers, from orchestration down to isolation

Agent Frameworks

7 tools

The middleware between a model and real actions — reasoning loops, state, tool execution, and multi-agent coordination.

Orchestration & Workflows

3 tools

Durable execution and workflow engines — retries, human-in-the-loop steps, and schedules for agent runs that outlive a single request.

AI Sandboxes

6 tools

Isolated compute where agent-generated code can run without putting the host at risk. MicroVM and container isolation.

Hosting & Deployment

3 tools

Where the agent actually runs — serverless functions, GPU capacity, and general-purpose cloud for the services around it.

Coding Agents & CLIs

7 tools

Terminal and IDE agents that read a codebase, propose multi-file changes, and iterate against a failing test suite.

MCP Infrastructure

4 tools

Registries, gateways, and scanners for the Model Context Protocol — how servers get discovered, deployed, and vetted.

Observability & Evals

5 tools

Tracing, token accounting, and evaluation. Standard APM misses the agent loop; these tools instrument the decisions.

Memory & State

3 tools

Persistence across sessions — vector recall, knowledge graphs, and stateful agent runtimes.

Data & Backend

9 tools

The databases and backend platforms agents read and write — Postgres hosts, serverless data, and app backends.

Vector & Retrieval

5 tools

Vector databases and embedding stores — the substrate underneath retrieval and long-term agent recall.

Auth & Governance

3 tools

Identity, delegated authorization, and policy for agents acting on a user's behalf.