☁ Hosted K8s Infrastructure

Your AI Agent.
One Pod. Full Sovereignty.

Launch an isolated Sea-Claw agent on our K3s cluster in under 10 seconds — or self-host on your own Kubernetes. 77 tools, 7 channels, arena memory, Grammar Shield. Your data stays in your pod.

K3s / K8s Native 77 Tools per Pod Agent Swarm HPA Autoscaling Open Source
terminal — seaclaw-platform
# Deploy in 3 commands
git clone github.com/t4tarzan/seabot
cd seabot/platform
./deploy.sh deploy
 
# Building images...
seaclaw-instance:latest built
seaclaw-gateway:latest built
Namespace seaclaw-platform created
Gateway deployed (port 30090)
 
# Sign up first agent:
POST /api/v1/agents/create
Pod: seaclaw-alice running
77 tools active | soul: alex
Discord + Telegram wired
<10s
Pod Spin-Up
77
Tools per Pod
7
Channels
1→5
HPA Replicas
5
Agent Souls
~17MB
Pod RAM
How It Works

From Signup to Running Agent in 10 Seconds

The gateway handles all K8s complexity. You just pick a model and soul.

01
🖊

Sign Up via UI or API

POST to /api/v1/agents/create with your username, LLM API key, preferred model, soul, and optional channel tokens. Everything else is automatic.

02

Gateway Spins a Pod

The gateway calls the K8s API to create an isolated pod, PVC, and service for your agent. Your API key is injected as a K8s env var — never baked into an image.

03
🤖

Agent Is Fully Live

Your pod starts the Sea-Claw binary with 77 tools, your chosen soul, and all channels wired. Talk to it via Telegram, Discord, API, or the chat UI immediately.

04
🔧

Manage via Dashboard

The gateway UI shows all running pods, lets you pause/restart agents, update the model, toggle souls, and stream live logs — all without touching kubectl.

05
🚀

Spawn a Swarm

Coordinator agents can call swarm_spawn to create ephemeral worker pods. Workers receive tasks via swarm_relay and are cleaned up automatically when done.

06
🔄

Scale Automatically

The gateway has an HPA configured (1→5 replicas). PodDisruptionBudget ensures zero downtime during rolling updates. LimitRange prevents any single pod from hogging resources.

Architecture

What Lives in the Cluster

A single K8s namespace containing everything — from the gateway RBAC to per-user PVCs.

seaclaw-platform namespace │ ├── gateway Deployment (FastAPI, 1–5 replicas via HPA) │ ├── ServiceAccount: gateway-sa │ ├── Role: gateway-role (create/delete pods+services+PVCs) │ ├── PVC: seaclaw-gateway-data (SQLite task DB, soul files) │ ├── Service: gateway-svc (ClusterIP :8090) │ └── Service: gateway-external (NodePort :30090) │ ├── seaclaw-<username> Pod (one per user, on-demand) │ ├── Container: sea_claw binary (~300KB, C11) │ │ ├── 77 tools (Git, PM, Swarm, Voice, File, Shell, Web, ...) │ │ ├── Grammar Shield (6-layer, <1µs/check) │ │ ├── Arena memory (16MB slab, zero malloc) │ │ └── 7 channels (Telegram, TUI, API, WS, Slack, Discord, Voice) │ ├── PVC: seaclaw-<username>-data (SQLite DB, workspace) │ └── Service: seaclaw-<username>-svc (ClusterIP :8899) │ ├── agent-zero Deployment (optional — Python multi-agent shim) │ ├── SeaZero shim (HTTP JSON-RPC, port 8080) │ ├── LLM Proxy (port 7432 — budget-controlled) │ └── Volumes: workdir, memory, shared │ └── Cluster-wide ├── HPA: gateway (min=1, max=5, targetCPU=70%) ├── PDB: gateway (minAvailable=1) ├── LimitRange: default container limits └── ResourceQuota: namespace CPU+RAM cap
🌐
Gateway (FastAPI)
Agent lifecycle, swarm management, config API, web UI, K8s API client
Always On
🤖
Agent Pods (C11)
One isolated pod per user. Own PVC, own env vars, own SQLite DB, own network policy
On Demand
🚀
Worker Pods (Swarm)
Ephemeral pods spawned by coordinator agents. Deleted when task completes or TTL expires
Ephemeral
🧠
Agent Zero (optional)
Python shim wrapping the Agent Zero autonomous agent framework. Budget-controlled via LLM proxy
Optional
🔒
RBAC & NetworkPolicy
Gateway ServiceAccount has minimal permissions. Each agent pod gets its own NetworkPolicy — only gateway can reach it
Always On
📊
HPA + PDB + Quotas
Gateway scales 1→5 replicas. PodDisruptionBudget ensures zero-downtime updates. ResourceQuota caps namespace usage
Auto
kubectl — seaclaw-platform
$ kubectl -n seaclaw-platform get pods
 
NAME READY STATUS
gateway-7d4f9c-xkp2r 1/1 Running
gateway-7d4f9c-nm8qt 1/1 Running
seaclaw-alice 1/1 Running
seaclaw-bob 1/1 Running
seaclaw-worker-1 1/1 Running
agent-zero-5b9f-rp4x2 1/1 Running
 
$ kubectl -n seaclaw-platform get svc
 
NAME TYPE PORT
gateway-svc ClusterIP 8090
gateway-external NodePort 30090
seaclaw-alice-svc ClusterIP 8899
seaclaw-bob-svc ClusterIP 8899
Features

Everything the Platform Gives You

Built on K3s. Designed for edge, homelab, and production alike.

Sub-10s Pod Startup

The Sea-Claw binary is ~300KB. Container startup is dominated by kubelet scheduling, not app init. Your agent is answering messages in under 10 seconds.

🔒

Full Pod Isolation

Each user gets their own K8s pod, PVC, and NetworkPolicy. No shared memory, no shared filesystem. Data is physically isolated at the container boundary.

🔧

77 Tools Built-In

File I/O, shell, web search, Git, project management, cron, DNS, SSL, JSON, CSV, math, voice transcription. All compiled into the binary — no plugins, no eval().

🧠

5 Agent Souls

Eva (analyst), Alex (developer), Tom (creative), Sarah (communicator), Max (generalist). Each soul shapes how the agent responds, prioritizes, and communicates.

📡

7 Input Channels

Telegram, Terminal, HTTP API, WebSocket, Slack, Discord, Voice (Whisper). Wire multiple channels to the same pod — same brain, same 77 tools, any interface.

🚀

Agent Swarm

Coordinator pods spawn ephemeral worker pods via swarm_spawn. Workers communicate via swarm_relay. The gateway manages the K8s lifecycle automatically.

📊

HPA Autoscaling

The gateway Deployment scales from 1 to 5 replicas when CPU hits 70%. PodDisruptionBudget ensures at least 1 replica stays up during rolling updates.

🔁

Multi-Node Ready

Run a single K3s node or a full multi-node cluster with NFS RWX storage. The setup-node.sh script joins worker nodes with correct role labels in one command.

🛠

One-Command Deploy

./deploy.sh deploy builds both Docker images, imports them into K3s containerd, applies all manifests in order, and waits for the gateway rollout. Done.

Channels

7 Ways to Talk to Your Pod

Configure channels at signup. All are optional except TUI — which always works.

💬

Telegram

Bot token + chat ID. Message your agent from anywhere.

💻

Terminal (TUI)

30+ slash commands. Always available inside the pod.

🌐

HTTP API

REST POST /api/chat. Integrate with any app.

🔌

WebSocket

Real-time streaming for dashboards and live UIs.

🗨

Slack

Outbound webhook. Agent responses go to your workspace.

🎮

Discord

Native bot polling. Set token + channel ID at signup.

🎤

Voice (Whisper)

Transcribe audio via Groq API or local whisper.cpp.

Gateway API

REST API for Agent Lifecycle

Everything the UI does, you can do via API. Automate agent creation, teardown, and config from your own tooling.

MethodEndpointDescription
POST/api/v1/agents/createSpawn a new agent pod
GET/api/v1/agentsList all running pods
GET/api/v1/agents/{user}Get pod status + info
DEL/api/v1/agents/{user}Delete pod + PVC
PATCH/api/v1/agents/{user}/configUpdate model, soul, swarm mode
POST/api/v1/agents/{user}/restartRestart pod
POST/api/v1/swarm/workersSpawn worker pod (swarm)
DEL/api/v1/swarm/workers/{name}Delete worker pod
POST/api/v1/swarm/relaySend message to worker agent
GET/healthGateway health check
example — create agent
POST /api/v1/agents/create
{
  "username": "alice",
  "openrouter_api_key": "sk-or-...",
  "model": "moonshotai/kimi-k2",
  "soul": "alex",
  "telegram_bot_token": "123:ABC...",
  "discord_bot_token": "MTxx..."
}
 
# 200 OK
{
  "status": "running",
  "pod": "seaclaw-alice",
  "service": "seaclaw-alice-svc",
  "tools": 77
}
Comparison

SeaClaw Platform vs Alternatives

How does our K8s agent platform compare to other ways of deploying AI agents?

Feature SeaClaw Platform ✦ Docker Compose Raw K8s YAML Managed AI (Cloud)
Deploy complexity 1 commanddocker compose up10+ manual stepsWeb UI signup
Data sovereignty 100% your cluster Local Local Cloud vendor
Multi-user isolation Pod per user~ Container per user Pod per user~ Shared infra
Built-in tools77Depends on imageDepends on image~20 (provider)
Agent Swarm K8s pods~ Manual
Autoscaling (gateway) HPA built-in~ Manual HPA Vendor handles
Channel support7 channelsSame (self-config)Same (self-config)1–3 channels
Edge / Raspberry Pi K3s <512MB Too heavy
Open source MIT
Get Started

Ready to Deploy?

Three commands. Fully self-hosted. Complete sovereignty over every pod.

Launch Agent → Deploy Guide View on GitHub