> ## Documentation Index
> Fetch the complete documentation index at: https://mezmo-9a59581a-promptless-aura-wait-for-tool.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# AURA: open source agent harness for production SRE work

> AURA is an open source agent harness for production SRE work. Scoped tools, approval gates, MCP-native, and a trace of every call.

<div style={{ display: 'flex', alignItems: 'center', gap: '8px', marginBottom: '16px' }}>
  <span style={{ display: 'inline-block', padding: '4px 10px', borderRadius: '999px', background: '#f3eefc', color: '#503A82', fontSize: '12px', fontWeight: 600 }}>
    Open source · Apache-2.0
  </span>
</div>

AURA is an open source agent harness for production operations work, written in Rust and licensed Apache 2.0. It takes an LLM plus the tools you already run and turns them into an agent that can work a live incident: reading cluster state, querying logs and metrics, checking your runbooks, and reporting what it found.

You declare the agent in TOML. Workers are separate roles, each with its own prompt and its own list of tools, so the worker that inspects a cluster is not the worker that changes it. Tools arrive over MCP, which means anything with an MCP server is reachable without writing an adapter.

## What it does

Model output alone does not make an agent you can point at production. The harness decides which tools each worker can reach, what pauses before a write, what happens when a tool call fails, what state survives across turns, and what record you have when the run is over.

AURA is that harness, and it is open source. You own the config and the deployment, and the telemetry goes where you point it. Nothing has to route through a vendor for the agent to work.

## Use cases

<CardGroup cols={2}>
  <Card title="Scope what the agent can reach" icon="bullseye" href="/aura/configuration-reference">
    Every worker gets an explicit tool list, so read access and write access live in different roles across the agent.
  </Card>

  <Card title="Stop before a destructive call" icon="octagon-exclamation" href="/aura/hitl">
    Approval gates name the mutating tools that need a human decision. Reads run straight through; writes wait for a yes.
  </Card>

  <Card title="Use the tools you already run" icon="plug" href="/aura/example-configs">
    MCP servers over HTTP, SSE, and stdio: Kubernetes, Prometheus, Grafana, GitHub, Notion, AWS, Mezmo, and anything else.
  </Card>

  <Card title="See what the agent did" icon="eye" href="/aura/tracing-spans">
    Every model call and tool call emits an OpenTelemetry span in OpenInference conventions, readable in any backend.
  </Card>

  <Card title="Answer from your runbooks" icon="book-open" href="/aura/skills">
    RAG over in-memory, Qdrant, or Bedrock knowledge bases, plus skills that load task instructions only when needed.
  </Card>

  <Card title="Run it where your data lives" icon="server" href="/aura/ollama-guide">
    OpenAI, Anthropic, Gemini, Bedrock, OpenRouter, or a local Ollama instance. Same config, different provider block.
  </Card>
</CardGroup>

## Start here

<CardGroup cols={3}>
  <Card title="Local quickstart" icon="terminal" href="/aura/quickstart">
    Install the binary, run `aura init`, connect an MCP server, and start asking questions.
  </Card>

  <Card title="Docker quickstart" icon="docker" href="/aura/quickstart-docker">
    Bring up AURA and a demo environment with Compose, no local installs required.
  </Card>

  <Card title="Kubernetes quickstart" icon="cubes" href="/aura/quickstart-k8s-sre">
    Run AURA as a long-lived in-cluster service with the Helm chart and a read-only Kubernetes MCP server.
  </Card>
</CardGroup>

## Examples

Each of these is a recorded run against a live environment. Where a video has a written walkthrough, the link goes straight to it.

### Install and connect

<CardGroup cols={2}>
  <Card title="Install AURA and debug an incident" icon="circle-play" href="https://www.mezmo.com/videos/install-aura-debug-incidents-with-an-open-source-sre-agent">
    Install on a desktop, run `aura init` against an Anthropic model, wire in a Grafana MCP server, and investigate.

    <span style={{ color: '#503A82', fontWeight: 600 }}>Watch the walkthrough →</span>
  </Card>

  <Card title="Install in Kubernetes with Helm" icon="circle-play" href="https://www.mezmo.com/videos/install-an-ai-sre-agent-in-kubernetes-with-aura-and-helm">
    Helm install AURA and the Kubernetes MCP server into a cluster, read-only through both RBAC and a startup flag.

    <span style={{ color: '#503A82', fontWeight: 600 }}>Watch the walkthrough →</span>
  </Card>

  <Card title="Connect MCP servers and scope tools" icon="circle-play">
    Add Mezmo, GitHub over HTTP streamable, and Grafana over stdio, then narrow which worker holds which tools.
  </Card>

  <Card title="Wire providers and pick a model" icon="circle-play">
    Point AURA at OpenAI, Anthropic, Gemini, Bedrock, OpenRouter, or a local Ollama with one provider block.
  </Card>
</CardGroup>

### Investigate an incident

<CardGroup cols={2}>
  <Card title="Find an unnoticed crash loop" icon="circle-play">
    A one-sentence health check turns up a pod in CrashLoopBackOff with 788 restarts that nobody asked about.
  </Card>

  <Card title="Turn a Slack report into a root cause" icon="circle-play">
    A coworker's message with no service name or time range. The log worker builds a window and traces checkout to payment.
  </Card>

  <Card title="Debug a Lambda timeout" icon="circle-play">
    AURA connects through the AWS MCP proxy with existing CLI credentials, finds the log group, and verifies the timeout.
  </Card>

  <Card title="Debug in Dash0 with no custom adapter" icon="circle-play">
    Two config sections point AURA at Dash0's hosted MCP server, read-only, with credentials in environment variables.
  </Card>

  <Card title="Debug Docker container permissions" icon="circle-play">
    The agent doesn't know the stack runs on Compose. It opens with generic advice, then works down to the mount and owner.
  </Card>

  <Card title="Trace a request across services" icon="circle-play">
    The tracing worker follows a slow request from the gateway through three services and back, no manual span-hunting.
  </Card>
</CardGroup>

### Give it your runbooks

<CardGroup cols={2}>
  <Card title="Read Notion runbooks during an incident" icon="circle-play">
    A read-only Notion key in `.env` and a runbook worker, so on-call guidance comes from your pages instead of the model.
  </Card>

  <Card title="Use GitHub as a runbook database" icon="circle-play">
    A fine-grained token scoped to one repo, plus sources of truth and routing rules for a runbook engineer worker.
  </Card>
</CardGroup>

### Keep it contained

<CardGroup cols={2}>
  <Card title="Add a human breakpoint" icon="circle-play">
    Deny a `pods_delete` call and confirm nothing changed, then approve the same call and watch it run.
  </Card>

  <Card title="Run entirely inside AWS with Bedrock" icon="circle-play">
    Bedrock as the provider, a managed knowledge base doing retrieval in-account, and every call landing in CloudTrail.
  </Card>
</CardGroup>

### Let it run on its own

<CardGroup cols={2}>
  <Card title="Schedule a Kubernetes guardian" icon="circle-play">
    A cron job starts a one-shot run with inspect access and exactly one mutation tool, bounded by prompt and by RBAC.
  </Card>

  <Card title="Fix a failed Argo CD deployment" icon="circle-play">
    A sync-fail hook sends a symptom over A2A. AURA finds the version mismatch and edits the one config map it may touch.
  </Card>

  <Card title="Go from log line to merged fix" icon="circle-play">
    A GitHub worker with a reduced tool list and an approval gate on issue creation, carrying a root cause into a PR.
  </Card>

  <Card title="Sweep for drift on a schedule" icon="circle-play">
    A nightly run compares live cluster state against declared config and opens an issue for each drift it finds.
  </Card>
</CardGroup>

## Join the project

<CardGroup cols={2}>
  <Card title="Join the AURA Slack" icon="slack" href="https://auracommunitygroup.slack.com/ssb/redirect">
    Ask questions, share what you are running, and talk through an approach before you open a PR.
  </Card>

  <Card title="Get started on GitHub" icon="github" href="https://github.com/mezmo/aura">
    Apache 2.0, written in Rust. Clone it, read the config examples, and open an issue when something does not fit.
  </Card>

  <Card title="Star the repo" icon="star" href="https://github.com/mezmo/aura">
    Stars are how other SREs find the project. If AURA saved you an afternoon, a star is the cheapest way to say so.
  </Card>

  <Card title="Have a use case in mind?" icon="lightbulb" href="https://www.mezmo.com/video-topics/aura">
    Tell us the tool, the failure you keep hitting, and what the agent should do. We build these into walkthroughs.
  </Card>
</CardGroup>