> ## Documentation Index
> Fetch the complete documentation index at: https://executioncontrolprotocol.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Execution Control Protocol

> The runtime specification for agentic systems

ECP (Execution Control Protocol) is an open standard for defining, packaging, versioning, and deploying **execution environments for AI agents** — portable specifications that describe what an agent can see, what tools it can access, and how it runs.

ECP is designed to **embrace and extend** the [Model Context Protocol (MCP)](https://modelcontextprotocol.io) — not replace it.

* **MCP** standardizes how models call tools.
* **ECP** standardizes how agents, tools, context, and policies execute together.

Think of ECP as:

* **Docker Compose for MCP tools** — bundle servers, mounts, and policies in one manifest
* **Infrastructure-as-code for agent execution** — declarative, versioned, shareable
* **The missing layer** between tool calling (MCP) and multi-agent systems

## What can ECP enable?

* **Shareable execution environments** — define once, run anywhere (CLI, platform, or custom runtime).
* **Verticalized contexts** — e.g. Shopify Ops Context, RevOps Context, with the right tools and guardrails.
* **Safe, inspectable execution** — default-deny tool access, budgets, write controls, and full audit traces.
* **Portable agent specs** — version and distribute Context manifests like container images.

## Why does ECP matter?

Depending on where you sit in the ecosystem, ECP can have different benefits.

* **Developers**: Define execution environments declaratively; no custom orchestration glue for every deployment.
* **Platforms and runtimes**: Implement once against the spec; support any ECP Context.
* **End users**: More capable, governed AI agents that run in reproducible, auditable environments.

## Relationship to MCP

ECP builds directly on MCP. MCP standardizes tool discovery and invocation; ECP does not redefine tool calling. Instead, ECP:

* References MCP servers and tool names
* Uses structured argument schemas
* Bundles permissions and policies
* Adds versioning and packaging semantics

If MCP is the "USB interface" for AI tools, ECP is the "container manifest" that defines the whole execution environment.

## Start building

<CardGroup cols={2}>
  <Card title="Quickstart" icon="rocket" href="/getting-started/quickstart">
    Run your first Context in minutes
  </Card>

  <Card title="Concepts" icon="book" href="/learn/concepts">
    Context, executors, orchestration, and mounts
  </Card>

  <Card title="Schema reference" icon="file-code" href="/reference/schema">
    JSON Schema and type definitions
  </Card>

  <Card title="CLI reference" icon="terminal" href="/reference/cli">
    ecp run, ecp validate, and ecp trace
  </Card>
</CardGroup>

## Learn more

<CardGroup cols={2}>
  <Card title="Relationship to MCP" icon="puzzle-piece" href="/learn/relationship-to-mcp">
    How ECP sits above MCP
  </Card>

  <Card title="Security" icon="shield" href="/learn/security">
    Default-deny, budgets, and audit
  </Card>
</CardGroup>

## Alternatives and comparisons

<CardGroup cols={2}>
  <Card title="Alternatives and comparisons" icon="scale-balanced" href="/alternatives">
    Compare ECP (Execution Control Protocol) to popular frameworks
  </Card>

  <Card title="ECP vs LangChain" icon="layers" href="/alternatives/ecp-vs-langchain">
    Orchestration vs execution control in practice
  </Card>
</CardGroup>

## AI tools

<CardGroup cols={2}>
  <Card title="Use Cursor with ECP" icon="arrow-pointer" href="/ai-tools/cursor">
    Install the ECP docs skill and move faster
  </Card>

  <Card title="Use Claude Code with ECP" icon="asterisk" href="/ai-tools/claude-code">
    Speed up manifests, examples, and CLI flows
  </Card>
</CardGroup>