llms.txt
Compare with other sites' llms.txt files →# Runta
> Agent execution layer for token efficiency, secret protection, and governed access.
Runta is a persistent agent runtime. It covers sandbox use cases and adds egress control, secrets, checkpoints, and audit.
Runta gives AI agents durable, isolated runtimes with controls for token spend, credentials, network access, and compute. Agents keep the tools and workflows they already use while Runta provides the execution boundary around them.
## Docs
- [Getting Started](https://runta.com/docs/): Create your first persistent runtime with the CLI, Python SDK, or TypeScript SDK, or hand setup to an agent.
- [Architecture](https://runta.com/docs/overview/): How Runta structures an agent runtime around files, egress control, and checkpoints.
- [Runtime Basics](https://runta.com/docs/runtime/runtime-basic/): Create, inspect, execute commands in, resize, pause, resume, stop, and delete Runta runtimes.
- [Token Saving](https://runta.com/docs/runtime/token-x-ray/): Find token-heavy tool calls with Token X-Ray and reduce supported tool output with Token Compression.
- [Auto Suspend and Wake-Up](https://runta.com/docs/runtime/auto-suspend-and-wake-up/): Suspend idle published runtimes and wake them when traffic returns.
- [Files](https://runta.com/docs/runtime/files/): Copy, upload, download, read, and write files in a Runta runtime.
- [Egress](https://runta.com/docs/runtime/egress/): Configure outbound allowlists and denylists for Runta runtimes.
- [Checkpoints](https://runta.com/docs/runtime/checkpoints/): Capture a runtime checkpoint and restore it into new Runta runtimes.
- [Secret Stubs](https://runta.com/docs/runtime/secrets-and-secret-injection/): Store tenant secrets and inject them into matching runtime outbound requests.
- [Publish a Service](https://runta.com/docs/runtime/publish-a-service/): Expose a runtime service over HTTPS through Runta ingress.
- [Memory Auto Scaling](https://runta.com/docs/runtime/memory-auto-scaling/): Start a runtime with a small memory request and let it grow up to a configured limit.
- [Team Access and Roles](https://runta.com/docs/organization/team-access-and-roles/): Invite collaborators, assign organization roles, and control what each member can do in Runta.
- [Cloud Agents](https://runta.com/docs/cloud-agents/): Managed AI agents with persistent cloud computers, conversations, streaming events, workspaces, and artifacts.
- [Cloud Agents Quickstart](https://runta.com/docs/cloud-agents/quickstart/): Create a Cloud Agent with a managed model provider, wait for its first run, read the reply, and clean up.
- [Cloud Agent Model Providers](https://runta.com/docs/cloud-agents/model-providers/): Discover managed providers, configure model routing, and use connected Codex subscriptions with Cloud Agents.
- [Cloud Agent Lifecycle](https://runta.com/docs/cloud-agents/lifecycle/): Create, list, rename, and delete agents, attach repositories, restore checkpoints, and share an existing cloud computer.
- [Cloud Agent Runs and Events](https://runta.com/docs/cloud-agents/runs-and-events/): Submit text and image prompts, follow up, resume, cancel, stream Pi events, and replay session events.
- [Cloud Agent Workspace and Artifacts](https://runta.com/docs/cloud-agents/workspace-and-artifacts/): Browse, search, read, write, and delete workspace files, inspect ports, and upload or download run artifacts.
- [Cloud Agent Computer Sessions](https://runta.com/docs/cloud-agents/computer-sessions/): Obtain an expiring VNC WebSocket ticket and connect a browser viewer to a Cloud Agent's desktop.
- [Cloud Agents API Reference](https://runta.com/docs/cloud-agents/api-reference/): Complete endpoint index, public schemas, authentication, limits, errors, and troubleshooting for the Cloud Agents REST API.
- [Agent Skills](https://runta.com/docs/reference/agent-skills/): Copy-ready Runta skills for installation, demos, the CLI, and SDKs.
- [Runta Installer Skill](https://runta.com/docs/reference/agent-skills/runta-installer/): Copy-ready SKILL.md content for agents installing Runta.
- [Runta Demos Skill](https://runta.com/docs/reference/agent-skills/runta-demos/): Copy-ready SKILL.md content for agents demonstrating Runta.
- [Runta CLI Skill](https://runta.com/docs/reference/agent-skills/runta-cli/): Copy-ready SKILL.md content for agents using the Runta CLI.
- [Runta Python SDK Skill](https://runta.com/docs/reference/agent-skills/runta-python-sdk/): Copy-ready SKILL.md content for agents using the Runta Python SDK.
- [Runta TypeScript SDK Skill](https://runta.com/docs/reference/agent-skills/runta-typescript-sdk/): Copy-ready SKILL.md content for agents using the Runta TypeScript SDK.
- [MCP Server](https://runta.com/docs/integrations/mcp/): Connect Claude, ChatGPT, Cursor, and other MCP clients to Runta.
- [Harbor](https://runta.com/docs/integrations/harbor/): Run Harbor jobs on the built-in Runta environment.
- [OpenAI Agents SDK](https://runta.com/docs/integrations/openai-agents/): Run OpenAI Agents SDK workflows on Runta runtimes, a persistent alternative to ephemeral sandboxes.
- [Runta CLI](https://runta.com/docs/reference/cli/): Command reference for the runta CLI.
- [Install and Configuration](https://runta.com/docs/reference/cli/install/): Install the runta CLI, configure your token, and verify it is available.
- [Runtime Commands](https://runta.com/docs/reference/cli/runtime-commands/): Create, list, inspect, resize, pause, resume, stop, and delete runtimes from the runta CLI.
- [Command Execution](https://runta.com/docs/reference/cli/exec/): Run commands inside a runtime with runta exec.
- [Files](https://runta.com/docs/reference/cli/files/): Copy files or directories between the local machine and a runtime with runta cp.
- [Checkpoints](https://runta.com/docs/reference/cli/checkpoints/): Manage runtime checkpoints through the runta checkpoint command group.
- [Egress](https://runta.com/docs/reference/cli/egress/): Manage runtime egress denylists and allowlists.
- [Secrets](https://runta.com/docs/reference/cli/secrets/): Store tenant secret values and manage secret stubs.
- [Create a Runta API Key](https://runta.com/docs/getting-started/cli-and-sdk-token/): Create a Runta API key and configure it for the CLI, Python SDK, and TypeScript SDK.
- [Credential Injection](https://runta.com/docs/getting-started/credential-injection/): Store credentials and inject them into matching outbound requests.
- [Governance and Egress](https://runta.com/docs/getting-started/governance-and-audit/): Configure runtime egress policy.
- [Memory Auto Scaling](https://runta.com/docs/getting-started/semantic-aware-auto-scaling/): Configure a runtime to grow from an initial memory request to a maximum limit.
- [Transfer Files](https://runta.com/docs/getting-started/transfer-files/): Copy files between the local machine and a Runta runtime.
- [Vertical Scaling](https://runta.com/docs/getting-started/vertical-scaling/): Resize runtime memory while keeping the workflow running.
- [Runta SDK Skills](https://runta.com/docs/reference/agent-skills/runta-sdk/): The combined Runta SDK skill has been split into separate Python and TypeScript SDK skills.
- [Runta SDK](https://runta.com/docs/reference/sdk/): Reference entry point for the Runta Python and TypeScript SDKs.
- [Agent Skills](https://runta.com/docs/skills/): Install Runta skills for coding agents.
## Pricing
- [Pricing](https://runta.com/pricing/): Usage-based Runta pricing for vCPU, memory, and persistent runtime storage.
## Security and compliance
SOC 2 Type II and ISO 27001 are in progress. Data residency is customer-selectable.
- [Trust Center](https://trust.inc/runta-inc): See the Runta Trust Center for SOC 2 and ISO 27001 status.
## API
- [Runta REST API](https://runta.com/docs/reference/api/): Generated API docs for runtimes, exec, checkpoints, files, egress, and secrets.
- [Public OpenAPI contract](https://runta.com/openapi.yaml): Public REST API for external developers and the runta CLI.
## SDKs
- [Python SDK Reference](https://runta.com/docs/reference/sdk/python/api/): Sync, Async, and Shared components exported by the Runta Python SDK.
- [Python SDK reference](https://runta.com/docs/reference/sdk/python/api/llms.txt): SDK version: 0.1.4. OpenAPI version: 3.0.3. Contract version: 0.0.5. Schema revision: sha256:14e1e000453ae6d3ea8d2a85bf75e7b9b443cdd492e7fc2d16d6eba57c96861c
- [Python SDK full reference](https://runta.com/docs/reference/sdk/python/api/llms-full.txt)
- [TypeScript SDK Reference](https://runta.com/docs/reference/sdk/typescript/api/): Promise-based TypeScript client, runtime, manager, model, and error reference.
- [TypeScript SDK reference](https://runta.com/docs/reference/sdk/typescript/api/llms.txt): SDK version: 0.0.4. OpenAPI version: 3.0.3. Contract version: 0.0.5. Schema revision: sha256:14e1e000453ae6d3ea8d2a85bf75e7b9b443cdd492e7fc2d16d6eba57c96861c
- [TypeScript SDK full reference](https://runta.com/docs/reference/sdk/typescript/api/llms-full.txt)
## MCP
- [Runta MCP server](https://mcp.runta.com/mcp): Streamable HTTP endpoint with OAuth sign-in for MCP clients.
- [Runta Claude connector](https://claude.ai/customize/connectors/id/8080e6e6-51ae-42cb-a0cd-1d10dcbdfbaf): Connect Claude to Runta.
- [Runta OpenAI plugin](https://chatgpt.com/plugins/plugin_asdk_app_6ad65dd9cb4c8191952bdfc913e7b31a): Connect ChatGPT to Runta.
## Blog
- [Blog: agents, harnesses, and runtimes](https://runta.com/blog/): Essays on agents, harnesses, and runtimes from the team building Runta's execution layer.
- [Grok Bot vs Muse: What Happens After Prompt Injection?](https://runta.com/blog/grok-bot-vs-muse-prompt-injection/): A look inside Grok Bot and Muse's cloud computers: how command review, network checks, and credential isolation limit what happens after prompt injection.
- [How We Built Errand with One Developer in One Week](https://runta.com/blog/building-errand-in-one-week/): I built Errand, an open-source alternative to Grok Bot and Muse, by myself in about a week. Confession: I didn't build the hard parts. Here's what Runta already had, what I wrote, and where the week actually went.
- [Introducing Errand: an open-source alternative to Grok Bot and Muse](https://runta.com/blog/introducing-errand/): Errand gives an AI agent its own computer to run tasks while you’re away—open-source alternative to Grok Bot and Muse.
- [Introducing Exo Harness for Runta Runtimes](https://runta.com/blog/introducing-exo-harness-for-runta-runtimes/): Exo Harness on a persistent Runta runtime, with protected model credentials and a workspace you can pause and resume.
- [Introducing FrontierHarness Eval](https://runta.com/blog/introducing-frontierharness-eval/): FrontierHarness eval is a harness benchmark of 9 production coding harnesses on 30 agentic tasks—same model and runtime. Pass rates cluster; cost spans 17x.
- [Introducing SSH access for Runta runtimes](https://runta.com/blog/ssh-access-for-runta-runtimes/): Connect any SSH-capable agent to a persistent Runta runtime, with access to services but without exposing the underlying credentials.
- [Agents aren't software. A conversation with Jeff Dean helped me see why.](https://runta.com/blog/agents-arent-software/): Reliability was designed at write time because code was deterministic. Agents break that assumption, and the layer beneath them, from checkpointing to recovery, has to be rebuilt around a new question: what does it mean to resume a half-finished judgment?
- [Runta: the execution layer for agents](https://runta.com/blog/runta-the-execution-layer-for-agents/): Runta is the execution layer for AI agents: computers they run on, secrets they can use without seeing, and access you govern. $20M seed led by Martin Casado at Andreessen Horowitz.
- [Introducing ClawShell, The Security Layer OpenClaw (Peter) Needs](https://runta.com/blog/introducing-clawshell/): ClawShell enforces OS-level isolation for AI agents — so even if the agent is fully compromised, it can't access what it was never given.
## Careers
- [Careers at Runta: build the agent execution layer](https://runta.com/careers/): We're hiring. Build the agent execution layer: runtimes, computers, and secrets for production agents.
- [Member of Technical Staff - Cloud Infrastructure](https://runta.com/careers/cloud-infrastructure/): Own the production cloud platform that makes Runta reliable, scalable, portable, and cost-efficient.
- [Member of Technical Staff - Core Runtime](https://runta.com/careers/core-runtime-engineer/): Build the execution engine that snapshots, resumes, branches, and recovers long-running AI agent processes.
- [Member of Technical Staff - Product Engineering](https://runta.com/careers/product-engineer/): Design the SDK, CLI, APIs, docs, and developer experience that make Runta's runtime primitives easy to use.
## Optional
- [Runta: The execution layer for AI agents](https://runta.com/): Runta cuts what agents waste, bounds what they can reach, and records everything they did.
- [About](https://runta.com/about/): Runta is the execution layer for AI agents: persistent runtimes, protected secrets, and control. Meet the team from Cloudflare and Kong.
- [Home](https://runta.com/index.md): The execution layer for AI agents.
- [Privacy policy](https://runta.com/privacy-policy/): Runta Privacy Notice, including how Runta collects, uses, and discloses information.
- [Request a demo](https://runta.com/request-demo/): Request a personalized demo of Runta's secure runtime infrastructure for AI agents.
- [Tags](https://runta.com/tags/): Topics covered on the Runta blog.
- [Terms of service](https://runta.com/terms-of-use/): Runta Terms of Service governing access to and use of Runta services.
- [Agent Skills index](https://runta.com/.well-known/agent-skills/index.json)