# Everruns > Everruns is open-source (MIT) infrastructure for AI agents: a headless, > durable agentic harness engine written in Rust and backed by PostgreSQL. > Agents run for hours or days, survive crashes, restarts, and worker loss, > and resume from persisted state rather than replaying from scratch. Everruns is API-first. Harnesses (the durable execution loop an agent runs on) are first-class, reusable objects, with agents as configuration on top. The operator console ships with the open-source core. Self-host the full stack, or use the managed edition at app.everruns.com. Every page on this site is also available as Markdown: request it with an `Accept: text/markdown` header. ## Start here - [Home](https://everruns.com/): Open-source (MIT) infrastructure for durable, long-running AI agents that run on infrastructure you own. - [Vision](https://everruns.com/vision/): Why Everruns exists and where the platform is going. - [Everruns Cloud](https://everruns.com/cloud/): The managed edition of the open-source platform, in early access at app.everruns.com, with built-in model access or your own provider keys. ## Products - [Everruns Platform](https://everruns.com/products/platform/): The headless, durable agentic harness engine. Define agents and tools, compose them into reusable harnesses, and ship them to Slack, web chat, scheduled jobs, webhooks, or a plain HTTP API. - [Everruns Framework](https://everruns.com/products/framework/): The application-facing everruns crate: agents, typed tools, multi-turn sessions, live events, and durable local state, running inside your own Rust binary with no services to operate. - [Yolop](https://everruns.com/yolop/): Minimalistic, hackable, forkable coding agent in Rust, built on the Everruns crates. Any model, including fully local ones, in the terminal or inside any ACP editor. - [Mira](https://everruns.com/mira/): A code-first eval framework for agents and tools. Write evals in Rust, Python, or TypeScript; one host CLI runs them across a model matrix, scores trajectories and budgets, and saves every run as files you commit. - [Bashkit](https://everruns.com/products/bashkit/): A virtual Bash interpreter with an in-process virtual filesystem, a sandbox for running untrusted scripts from AI agents and users, with no containers or subprocesses. ## Platform concepts - [Capabilities](https://everruns.com/platform/capabilities/): The registry of tools and behaviors agents act through: execution, browser, network, storage, UI, and session control. - [Integrations](https://everruns.com/platform/integrations/): Execution sandboxes, evaluation tooling, model interfaces, and MCP bridges that connect Everruns to the rest of a stack. - [LLM Providers](https://everruns.com/platform/providers/): The model providers Everruns can route to, and how provider credentials are managed. - [Sandboxes](https://everruns.com/platform/sandboxes/): Isolated execution environments for agent code, local and cloud-hosted. ## Use cases - [Long-Running Research Agents](https://everruns.com/use-cases/long-running-research-agents/): Run long-running AI research agents for hours or days with durable execution, persisted progress, and automatic recovery from crashes, restarts, or worker loss. - [Headless Agent Platform](https://everruns.com/use-cases/headless-agent-platform/): Run headless AI agents as backend services with API integration, centralized governance, durable execution, and control inside your security perimeter. - [Physical World Agents](https://everruns.com/use-cases/physical-world-agents/): Orchestrate physical-world AI agents for labs, sensors, and robotics with durable execution that keeps extended workflows recoverable after failures at scale. ## Comparisons - [Everruns vs Claude Managed Agents](https://everruns.com/compare/everruns-vs-claude-managed-agents/): Everruns vs Claude Managed Agents: a self-hostable, provider-neutral, MIT-licensed durable agent platform compared with Anthropic’s hosted, Claude-only managed agents. - [Everruns vs LangGraph Platform](https://everruns.com/compare/everruns-vs-langgraph-platform/): Everruns vs LangGraph Platform (LangSmith Deployment): compare a fully MIT, self-hostable durable agent platform with LangChain’s production agent runtime and its pricing. - [Everruns vs Mastra](https://everruns.com/compare/everruns-vs-mastra/): Everruns vs Mastra: compare a Rust + PostgreSQL durable agent platform with no wall-clock ceiling against Mastra’s TypeScript-first, serverless, open-core agent framework. ## Blog - [Introducing the Agent API: A Stable, Unified Way to Invoke Agents](https://everruns.com/blog/introducing-agent-api/): Every Everruns agent now has one stable, unified API: the same URL, keys, routes and SDK client on Everruns Cloud, a self-hosted server, or serve. - [September 2026: Everruns Becomes a Framework, What's New](https://everruns.com/blog/whats-new-september-2026/): What shipped across Everruns since the July roundup, the everruns crate and the Everruns Framework, typed decisions with Jev, Cloud opening, and Ask User. - [July 2026: Two New Projects and a Cloud Taking Shape, What's New](https://everruns.com/blog/whats-new-july-2026/): What shipped across Everruns since the June roundup, two new open-source projects, Yolop and Mira, and platform work paving the way to Everruns Cloud. - [June 2026: A Month of Everruns, What's New](https://everruns.com/blog/whats-new-spring-2026/): A roundup of what shipped across Everruns in June 2026, knowledge indexes, resource discovery, durable subagents, tool-call repair, guardrails, judges, and more. - [Embedding the Everruns Runtime: A Local Coding Agent in One Binary](https://everruns.com/blog/embedding-the-runtime/): Most of what we publish about Everruns is the server side. But the same runtime powers in-process agents too. Here is what embedding looks like, using a terminal coding agent as the worked example. - [Request Signing: Cryptographic Identity for AI Agents](https://everruns.com/blog/request-signing/): When an AI agent browses the web on your behalf, how does the target server know who sent it? Request signing replaces guesswork with cryptographic proof. - [Introducing Everruns: Durable Infrastructure for AI Agents](https://everruns.com/blog/introducing-everruns/): Introducing Everruns, an open source infrastructure layer for durable AI agents that run for hours or days and recover from crashes, restarts, and failures. ## Workflow skills - [everruns](https://everruns.com/.well-known/agent-skills/everruns/SKILL.md): Understand Everruns products, core concepts, and where to find documentation. Use for product questions, getting started, or choosing between Cloud, the self-hosted Platform, and the Rust Framework. - [everruns-cli](https://everruns.com/.well-known/agent-skills/everruns-cli/SKILL.md): Use the everruns command tree to inspect and manage agents, sessions, integrations, triggers, and other platform resources. Use for CLI usage, shell scripting, and Platform Chat operations; use everruns-agent-authoring for agent design. - [everruns-mcp](https://everruns.com/.well-known/agent-skills/everruns-mcp/SKILL.md): Connect to and use the Everruns MCP endpoint to find agents, run and continue sessions, inspect results, and manage platform resources. Use for MCP connection setup and work through an Everruns MCP server; for authoring agents, also load everruns-agent-authoring. Use everruns-cli for command usage and examples, everruns for product concepts, and everruns-framework for Rust development. - [everruns-agent-authoring](https://everruns.com/.well-known/agent-skills/everruns-agent-authoring/SKILL.md): Create or revise Everruns platform agents through MCP, including instructions, harnesses, capabilities, model selection, portable packages, and behavioral verification. Use when changing an agent definition; use everruns-mcp for running an existing agent or inspecting sessions. - [everruns-framework](https://everruns.com/.well-known/agent-skills/everruns-framework/SKILL.md): Build Rust applications with the Everruns Framework or compose advanced execution hosts with everruns_core::host. Use for in-process agents, providers, tools, scoped MCP clients, and capability authoring. For product concepts use everruns; for platform operations use everruns-cli or everruns-mcp, and for agent definitions use everruns-agent-authoring. - [everruns-plugin](https://everruns.com/.well-known/agent-skills/everruns-plugin/SKILL.md): Install the Everruns plugin in Claude Code, Codex, Cursor, or Gemini CLI so a coding agent can build, run, debug, and ship Everruns agents. Use when a person wants their coding agent to work with Everruns. - [everruns-api](https://everruns.com/.well-known/agent-skills/everruns-api/SKILL.md): Create agents, open sessions, and stream events over the Everruns REST API with a personal access token. Use for server-to-server automation where no browser is available. - [everruns-self-host](https://everruns.com/.well-known/agent-skills/everruns-self-host/SKILL.md): Run the full open-source Everruns platform locally or on your own infrastructure with Docker Compose. Use when evaluating Everruns or deploying it inside your own perimeter. ## Optional - [llms-full.txt](https://everruns.com/llms-full.txt): every page on this site, concatenated as one document. - [Documentation](https://everruns.com/docs/): installation, architecture, API reference, and operations. - [Documentation llms.txt](https://everruns.com/docs/llms.txt): the docs index for language models, with per-topic text sets and the OpenAPI schema. - [Source](https://github.com/everruns/everruns): the MIT-licensed platform repository. - [auth.md](https://everruns.com/auth.md): how an agent obtains credentials for the Everruns API. - [Everruns plugin](https://github.com/everruns/plugins): install in Claude Code, Codex, Cursor, or Gemini CLI so a coding agent can build, run, debug, and ship Everruns agents. - [Coding agent setup](https://everruns.com/agent-setup/prompt.md): connect your coding agent using the official plugin instructions, browser sign-in and a read-only connection check.