Runloop
Runloop is the AI Agent Accelerator — secure code sandboxes (Devboxes), evaluation infrastructure (Benchmarks, Scenarios), and production-grade orchestration for AI coding agents at enterprise scale. The platform provides a single REST API and matched Python/TypeScript SDKs covering Devbox lifecycle, Blueprint image building, Snapshot branching, Agent registry, Axon event streams with Broker bridges, Storage Objects, Secrets, Network Policies, Agent Gateways, MCP Configs, and a turnkey eval framework (SWE-Bench Verified, SWE-smith, custom benchmarks). Runloop runs on a custom bare-metal hypervisor with microVM-level isolation and is SOC 2 Type II, HIPAA, and GDPR compliant, with VPC deployment available for regulated workloads. Founded 2023 by Jonathan Wall (Google File System lead, Google Wallet co-founder, Index CTO acquired by Stripe).
13 APIs
0 Features
AIAI AgentsCoding AgentsSandboxesDevboxesCode ExecutionEvaluationBenchmarksSWE-BenchMCPSnapshotsmicroVMEnterpriseSOC 2
Create, configure, and operate Devboxes — isolated Linux microVM-backed sandbox environments purpose-built for AI coding agents. The Devbox API covers the full lifecycle (create...
Build, preview, list, and manage Blueprints — Dockerfile-based container image templates that Devboxes launch from. Blueprints capture pre-installed system packages, language ru...
Define, configure, and run Benchmarks against your agents. Runloop ships SWE-Bench Verified and SWE-smith out of the box; the Benchmark API also supports custom benchmarks built...
Author and run individual evaluation Scenarios — the atomic test unit for AI coding agents. Each Scenario captures input context, environment setup, an agent invocation, and one...
Register and version Agents — packaged agent definitions sourced from Git, npm, pip, or pre-packaged storage objects. Once registered, an Agent can be mounted to any Devbox at c...
Create and operate Axons — Runloop's distributed event streams for sequencing, recording, and observing agent interactions. Each Axon supports event publish, SSE subscription, S...
Upload and manage Storage Objects — managed files with presigned download URLs that can be mounted into Devboxes or shared between agent runs. Two-step upload (create -> complet...
Securely manage account-level Secrets (API keys, tokens, credentials) that Runloop injects into Devboxes at runtime. Raw secret values are never exposed to agent code or to the ...
Define and manage egress Network Policies that restrict outbound network access from Devboxes. Allow/deny rules at account scope, attachable per Devbox at launch.
Configure Agent Gateways — proxy configurations that let Devbox-resident agents call external AI provider APIs (OpenAI, Anthropic, etc.) without exposing the raw provider creden...
Configure and manage Model Context Protocol (MCP) server bindings for Devboxes. Each MCP Config is a declarative mount that Runloop wires into Devbox agent runtimes so the agent...
Create and manage Runloop account API keys and Restricted Keys (scoped, narrower-permission tokens). API keys are the standard auth credential for SDK and REST API access; Restr...
Stream stdout and stderr updates from in-flight Devbox executions in real time. Use alongside the Devbox execute endpoints for long-running command output capture.
aid: runloop-ai
url: https://raw.githubusercontent.com/api-evangelist/runloop-ai/refs/heads/main/apis.yml
apis:
- aid: runloop-ai:runloop-devbox-api
name: Runloop Devbox API
tags:
- AI
- AI Agents
- Sandboxes
- Devboxes
- Coding Agents
humanURL: https://docs.runloop.ai/docs/devboxes/overview
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/docs/devboxes/overview
type: Documentation
- url: https://docs.runloop.ai/docs/devboxes/lifecycle
type: Documentation
- url: https://docs.runloop.ai/docs/devboxes/execute-commands
type: Documentation
- url: https://docs.runloop.ai/docs/devboxes/files
type: Documentation
- url: https://docs.runloop.ai/docs/devboxes/named-shells
type: Documentation
- url: https://docs.runloop.ai/docs/devboxes/snapshots
type: Documentation
- url: https://docs.runloop.ai/docs/devboxes/start-stop
type: Documentation
- url: https://docs.runloop.ai/docs/devboxes/tunnels
type: Documentation
- url: openapi/runloop-devbox-api-openapi.yml
type: OpenAPI
- url: json-schema/runloop-devbox-schema.json
type: JSONSchema
- url: json-schema/runloop-execution-schema.json
type: JSONSchema
- url: json-schema/runloop-snapshot-schema.json
type: JSONSchema
- url: json-schema/runloop-tunnel-schema.json
type: JSONSchema
- url: json-schema/runloop-launch-parameters-schema.json
type: JSONSchema
- url: json-structure/runloop-devbox-structure.json
type: JSONStructure
- url: json-ld/runloop-context.jsonld
type: JSONLD
- url: examples/runloop-devbox-create-example.json
type: Example
- url: examples/runloop-devbox-exec-example.json
type: Example
- url: examples/runloop-snapshot-create-example.json
type: Example
description: >
Create, configure, and operate Devboxes — isolated Linux microVM-backed sandbox environments purpose-built for AI
coding agents. The Devbox API covers the full lifecycle (create, suspend, resume, snapshot, shutdown), synchronous
and asynchronous command execution, named shell sessions, file I/O, code-repo mounts, agent and storage object
mounts, network tunnels with HTTPS URLs, SSH key issuance, and rich observability (executions, metrics, logs).
Includes the PTY control-plane for interactive WebSocket shell sessions.
- aid: runloop-ai:runloop-blueprint-api
name: Runloop Blueprint API
tags:
- AI
- AI Agents
- Blueprints
- Container Images
- Devboxes
humanURL: https://docs.runloop.ai/docs/devboxes/blueprints/overview
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/docs/devboxes/blueprints/overview
type: Documentation
- url: openapi/runloop-blueprint-api-openapi.yml
type: OpenAPI
- url: json-schema/runloop-blueprint-schema.json
type: JSONSchema
description: >
Build, preview, list, and manage Blueprints — Dockerfile-based container image templates that Devboxes launch
from. Blueprints capture pre-installed system packages, language runtimes, code mounts, ports, and agent tooling
so a fresh Devbox starts in seconds. Supports private blueprints, public discovery, metadata-keyed search, and
creation from RepositoryConnection inspection.
- aid: runloop-ai:runloop-benchmark-api
name: Runloop Benchmark API
tags:
- AI
- AI Agents
- Benchmarks
- Evaluation
- SWE-Bench
humanURL: https://docs.runloop.ai/docs/benchmarks/overview
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/docs/benchmarks/overview
type: Documentation
- url: openapi/runloop-benchmark-api-openapi.yml
type: OpenAPI
- url: json-schema/runloop-benchmark-schema.json
type: JSONSchema
- url: json-schema/runloop-benchmark-run-schema.json
type: JSONSchema
- url: examples/runloop-benchmark-run-example.json
type: Example
description: >
Define, configure, and run Benchmarks against your agents. Runloop ships SWE-Bench Verified and SWE-smith out of
the box; the Benchmark API also supports custom benchmarks built from your own scenarios and scorers. Resources
include benchmarks, benchmark runs (with start, cancel, complete lifecycle), benchmark jobs, scenario runs, and
downloadable run logs.
- aid: runloop-ai:runloop-scenario-api
name: Runloop Scenario API
tags:
- AI
- AI Agents
- Scenarios
- Evaluation
humanURL: https://docs.runloop.ai/docs/benchmarks/scenarios
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/docs/benchmarks/overview
type: Documentation
- url: openapi/runloop-scenario-api-openapi.yml
type: OpenAPI
- url: json-schema/runloop-scenario-schema.json
type: JSONSchema
description: >
Author and run individual evaluation Scenarios — the atomic test unit for AI coding agents. Each Scenario captures
input context, environment setup, an agent invocation, and one or more Scenario Scorers that produce pass/fail
signals plus structured scores. Used standalone or composed into Benchmarks.
- aid: runloop-ai:runloop-agents-api
name: Runloop Agents API
tags:
- AI
- AI Agents
- Agent Registry
humanURL: https://docs.runloop.ai/docs/devboxes/agents/using-agents-api
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/docs/devboxes/agents/using-agents-api
type: Documentation
- url: openapi/runloop-agents-api-openapi.yml
type: OpenAPI
- url: json-schema/runloop-agent-schema.json
type: JSONSchema
description: >
Register and version Agents — packaged agent definitions sourced from Git, npm, pip, or pre-packaged storage
objects. Once registered, an Agent can be mounted to any Devbox at create time for fast, reproducible installs.
Includes public agent listings and per-agent Devbox-count reporting.
- aid: runloop-ai:runloop-axons-api
name: Runloop Axons API
tags:
- AI
- AI Agents
- Event Streams
- Axon
- Broker
- ACP
humanURL: https://docs.runloop.ai/docs/axons/overview
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/docs/axons/overview
type: Documentation
- url: https://docs.runloop.ai/docs/axons/broker
type: Documentation
- url: openapi/runloop-axons-api-openapi.yml
type: OpenAPI
- url: json-schema/runloop-axon-schema.json
type: JSONSchema
- url: examples/runloop-axon-subscribe-example.json
type: Example
description: >
Create and operate Axons — Runloop's distributed event streams for sequencing, recording, and observing agent
interactions. Each Axon supports event publish, SSE subscription, SQL queries and batch operations over event
history, and a Broker bridge that connects an Axon to agents running inside Devboxes using ACP or Claude agent
client protocols.
- aid: runloop-ai:runloop-objects-api
name: Runloop Storage Objects API
tags:
- AI
- AI Agents
- Storage
- Objects
humanURL: https://docs.runloop.ai/docs/storage-objects/overview
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/docs/storage-objects/overview
type: Documentation
- url: openapi/runloop-objects-api-openapi.yml
type: OpenAPI
- url: json-schema/runloop-object-schema.json
type: JSONSchema
description: >
Upload and manage Storage Objects — managed files with presigned download URLs that can be mounted into Devboxes
or shared between agent runs. Two-step upload (create -> complete), public listings, metadata-keyed search, and
download presign.
- aid: runloop-ai:runloop-secrets-api
name: Runloop Secrets API
tags:
- AI
- AI Agents
- Secrets
- Credentials
humanURL: https://docs.runloop.ai/docs/devboxes/configuration/account-secrets
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/docs/devboxes/configuration/account-secrets
type: Documentation
- url: openapi/runloop-secrets-api-openapi.yml
type: OpenAPI
- url: json-schema/runloop-secret-schema.json
type: JSONSchema
description: >
Securely manage account-level Secrets (API keys, tokens, credentials) that Runloop injects into Devboxes at
runtime. Raw secret values are never exposed to agent code or to the Devbox shell environment after creation —
they are referenced by name.
- aid: runloop-ai:runloop-network-policies-api
name: Runloop Network Policies API
tags:
- AI
- AI Agents
- Network Policy
- Security
humanURL: https://docs.runloop.ai/docs/network-policies
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/docs/network-policies
type: Documentation
- url: openapi/runloop-network-policies-api-openapi.yml
type: OpenAPI
- url: json-schema/runloop-network-policy-schema.json
type: JSONSchema
description: >
Define and manage egress Network Policies that restrict outbound network access from Devboxes. Allow/deny rules at
account scope, attachable per Devbox at launch.
- aid: runloop-ai:runloop-gateway-configs-api
name: Runloop Gateway Configs API
tags:
- AI
- AI Agents
- Agent Gateways
- Credential Proxy
humanURL: https://docs.runloop.ai/docs/devboxes/agent-gateways
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/docs/devboxes/agent-gateways
type: Documentation
- url: openapi/runloop-gateway-configs-api-openapi.yml
type: OpenAPI
- url: json-schema/runloop-gateway-config-schema.json
type: JSONSchema
description: >
Configure Agent Gateways — proxy configurations that let Devbox-resident agents call external AI provider APIs
(OpenAI, Anthropic, etc.) without exposing the raw provider credentials to agent code. Implements the Credential
Gateway / opaque token-injection security pattern.
- aid: runloop-ai:runloop-mcp-configs-api
name: Runloop MCP Configs API
tags:
- AI
- AI Agents
- MCP
- Model Context Protocol
humanURL: https://docs.runloop.ai/docs/devboxes/mcp
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/docs/tools/ai-tools
type: Documentation
- url: openapi/runloop-mcp-configs-api-openapi.yml
type: OpenAPI
- url: json-schema/runloop-mcp-config-schema.json
type: JSONSchema
description: >
Configure and manage Model Context Protocol (MCP) server bindings for Devboxes. Each MCP Config is a declarative
mount that Runloop wires into Devbox agent runtimes so the agent can call MCP tools without bespoke setup.
- aid: runloop-ai:runloop-apikeys-api
name: Runloop API Keys API
tags:
- AI
- AI Agents
- API Keys
- Authentication
humanURL: https://docs.runloop.ai/api-reference/apikeys/create-api-key
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/api-reference/apikeys/create-api-key
type: Documentation
- url: openapi/runloop-apikeys-api-openapi.yml
type: OpenAPI
- url: json-schema/runloop-api-key-schema.json
type: JSONSchema
- url: json-schema/runloop-restricted-key-schema.json
type: JSONSchema
description: >
Create and manage Runloop account API keys and Restricted Keys (scoped, narrower-permission tokens). API keys are
the standard auth credential for SDK and REST API access; Restricted Keys are intended for embedded use in
Devbox-resident agents.
- aid: runloop-ai:runloop-executions-api
name: Runloop Executions Streaming API
tags:
- AI
- AI Agents
- Executions
- Streaming
humanURL: https://docs.runloop.ai/docs/devboxes/execution-logs
baseURL: https://api.runloop.ai
properties:
- url: https://docs.runloop.ai/docs/devboxes/execution-logs
type: Documentation
- url: openapi/runloop-executions-api-openapi.yml
type: OpenAPI
description: >
Stream stdout and stderr updates from in-flight Devbox executions in real time. Use alongside the Devbox execute
endpoints for long-running command output capture.
name: Runloop
description: >
Runloop is the AI Agent Accelerator — secure code sandboxes (Devboxes), evaluation infrastructure (Benchmarks,
Scenarios), and production-grade orchestration for AI coding agents at enterprise scale. The platform provides a
single REST API and matched Python/TypeScript SDKs covering Devbox lifecycle, Blueprint image building, Snapshot
branching, Agent registry, Axon event streams with Broker bridges, Storage Objects, Secrets, Network Policies, Agent
Gateways, MCP Configs, and a turnkey eval framework (SWE-Bench Verified, SWE-smith, custom benchmarks). Runloop runs
on a custom bare-metal hypervisor with microVM-level isolation and is SOC 2 Type II, HIPAA, and GDPR compliant, with
VPC deployment available for regulated workloads. Founded 2023 by Jonathan Wall (Google File System lead, Google
Wallet co-founder, Index CTO acquired by Stripe).
tags:
- AI
- AI Agents
- Coding Agents
- Sandboxes
- Devboxes
- Code Execution
- Evaluation
- Benchmarks
- SWE-Bench
- MCP
- Snapshots
- microVM
- Enterprise
- SOC 2
kind: contract
image: https://kinlane-images.s3.amazonaws.com/shared/apis-json/apis-json-logo.jpg
access: 3rd-Party
created: '2026-05-25'
modified: '2026-05-25'
common:
- type: PostmanWorkspace
url: https://www.postman.com/kinlaneapi/runloop/overview
- type: ArazzoWorkflows
url: arazzo/
workflows:
- url: arazzo/runloop-ai-bootstrap-devbox-from-object-workflow.yml
name: Runloop Bootstrap a Devbox from an Object Artifact
summary: >-
Create and complete a storage object, generate its download URL, then boot a devbox and pull the artifact into
it.
- url: arazzo/runloop-ai-build-blueprint-and-launch-devbox-workflow.yml
name: Runloop Build Blueprint and Launch Devbox
summary: >-
Build a custom blueprint image, poll until the build completes, then launch a devbox from it and wait for
running.
- url: arazzo/runloop-ai-create-scenario-and-run-workflow.yml
name: Runloop Create Scenario and Run It
summary: >-
Define a repeatable AI coding evaluation scenario, start a run of it, poll until it is scored, then complete
the run.
- url: arazzo/runloop-ai-graceful-snapshot-and-shutdown-workflow.yml
name: Runloop Graceful Snapshot and Shutdown
summary: Synchronously snapshot a running devbox's disk to preserve its state, then permanently shut the devbox down.
- url: arazzo/runloop-ai-provision-and-run-command-workflow.yml
name: Runloop Provision Devbox and Run a Command
summary: >-
Create a devbox, wait for it to reach running, then execute a shell command asynchronously and wait for
completion.
- url: arazzo/runloop-ai-publish-object-artifact-workflow.yml
name: Runloop Publish an Object Artifact
summary: >-
Create a storage object, mark its upload complete to make it read-only, then generate a presigned download URL
for it.
- url: arazzo/runloop-ai-score-running-scenario-run-workflow.yml
name: Runloop Score a Running Scenario Run
summary: Start a run of an existing scenario, wait for it to be running, trigger scoring, then complete the run.
- url: arazzo/runloop-ai-snapshot-and-restore-devbox-workflow.yml
name: Runloop Snapshot and Restore a Devbox
summary: >-
Take an asynchronous disk snapshot of a running devbox, poll until it completes, then launch a new devbox from
that snapshot.
- url: arazzo/runloop-ai-suspend-and-resume-devbox-workflow.yml
name: Runloop Suspend and Resume a Devbox
summary: >-
Suspend a running devbox to free compute, wait until it is suspended, then resume it and wait until it is
running again.
- url: arazzo/runloop-ai-write-execute-read-file-workflow.yml
name: Runloop Write, Execute, and Read a File on a Devbox
summary: >-
Boot a devbox, write a file into it, run a command that transforms the file, then read the resulting contents
back.
- type: Website
url: https://runloop.ai
- type: Documentation
url: https://docs.runloop.ai
- type: APIReference
url: https://docs.runloop.ai/api-reference
- type: Portal
url: https://docs.runloop.ai
- type: SignUp
url: https://platform.runloop.ai/signup
- type: Pricing
url: https://runloop.ai/pricing
- type: Plans
url: plans/runloop-ai-plans-pricing.yml
- type: RateLimits
url: rate-limits/runloop-ai-rate-limits.yml
- type: FinOps
url: finops/runloop-ai-finops.yml
- type: Vocabulary
url: vocabulary/runloop-ai-vocabulary.yml
- type: SpectralRules
url: rules/runloop-ai-rules.yml
- type: JSONLD
url: json-ld/runloop-context.jsonld
- type: StatusPage
url: https://status.runloop.ai
- type: Security
url: https://runloop.ai/security
- type: Compliance
url: https://runloop.ai/security
- type: Careers
url: https://runloop.ai/careers
- type: About
url: https://runloop.ai/about
- type: Blog
url: https://runloop.ai/blog
- type: Media
url: https://runloop.ai/in-the-media
- type: ContactSales
url: https://runloop.ai/contact
- type: GitHub
url: https://github.com/runloopai
- type: Documentation
name: llms.txt
url: https://docs.runloop.ai/llms.txt
- type: Documentation
name: Quickstart
url: https://docs.runloop.ai/docs/tutorials/quickstart
- type: Documentation
name: Features
url: https://docs.runloop.ai/docs/overview/runloop-features
- type: Documentation
name: What is Runloop?
url: https://docs.runloop.ai/docs/overview/what-is-runloop
- type: Tutorials
url: https://docs.runloop.ai/docs/tutorials/overview
- type: SDK
name: Python SDK
url: https://github.com/runloopai/api-client-python
- type: SDK
name: Python SDK Docs
url: https://runloopai.github.io/api-client-python/
- type: PythonPackage
name: runloop-api-client
url: https://pypi.org/project/runloop-api-client/
- type: SDK
name: TypeScript SDK
url: https://github.com/runloopai/api-client-ts
- type: SDK
name: TypeScript SDK Docs
url: https://runloopai.github.io/api-client-ts/
- type: NPMPackage
name: '@runloop/api-client'
url: https://www.npmjs.com/package/@runloop/api-client
- type: CLI
name: rl-cli
url: https://github.com/runloopai/rl-cli
- type: CLI
name: rl-cli docs
url: https://docs.runloop.ai/docs/tools/rl-cli
- type: Homebrew
url: https://github.com/runloopai/homebrew-tap
- type: SDK
name: Remote Agents SDK (Axon broker clients)
url: https://github.com/runloopai/remote-agents-sdk
- type: Tools
name: Runloop Examples (cookbook)
url: https://github.com/runloopai/runloop-examples
- type: Tools
name: deploy-agent GitHub Action
url: https://github.com/runloopai/deploy-agent
- type: Tools
name: code-execution-agent
url: https://github.com/runloopai/code-execution-agent
- type: Tools
name: AI Agent App Template
url: https://github.com/runloopai/ai-agent-app-template
- type: Tools
name: Cursor Rules (Python)
url: https://docs.runloop.ai/static/files/runloop-python-client.mdc
- type: Tools
name: Cursor Rules (TypeScript)
url: https://docs.runloop.ai/static/files/runloop-typescript-client.mdc
- type: Tools
name: ComputeSDK (open-source code execution toolkit)
url: https://github.com/runloopai/computesdk
- type: Tools
name: deepagents
url: https://github.com/runloopai/deepagents
- type: Tools
name: codex-tax-man (Codex + W&B Weave starter)
url: https://github.com/runloopai/codex-tax-man
- type: Integration
name: OpenAI Agents SDK + Runloop tutorial
url: https://docs.runloop.ai/docs/tutorials/openai-agentssdk-runloop
- type: Integration
name: OpenCode on Runloop
url: https://docs.runloop.ai/docs/tutorials/opencode-runloop
- type: Integration
name: Axon + ACP with OpenCode
url: https://docs.runloop.ai/docs/tutorials/axon-acp-broker
- type: Integration
name: GitHub Actions for agent deployment
url: https://docs.runloop.ai/docs/devboxes/agents/deploying-with-github-actions
- type: MCP
name: Runloop MCP Server / AI Tools
url: https://docs.runloop.ai/docs/tools/ai-tools
- type: LinkedIn
url: https://www.linkedin.com/company/runloopai
- type: Twitter
url: https://twitter.com/runloopai
- type: ContactInfo
name: support@runloop.ai
url: mailto:support@runloop.ai
maintainers:
- FN: Runloop, Inc.
url: https://runloop.ai
email: support@runloop.ai