blissful-infra dashboard
The dashboard is a single control plane for every tenant on the machine. It runs in its own container on localhost:3002, separate from any tenant’s network.
blissful-infra dashboard upSubcommands
Section titled “Subcommands”| Command | What it does |
|---|---|
dashboard up | Start the dashboard and open it in your browser |
dashboard down | Stop the dashboard |
dashboard status | Show whether it is running |
dashboard open | Open the running dashboard in your browser |
dashboard login | Authenticate the Agent tab with Claude (one-time, no API key needed) |
dashboard up takes --no-open to skip launching the browser.
One dashboard, every tenant
Section titled “One dashboard, every tenant”The dashboard is host-level, not per-tenant. A tenant switcher in the header re-scopes the whole UI, and every request it makes carries the selected tenant. You do not run one dashboard per tenant.
| Tab | What it does |
|---|---|
| Logs | Live log streaming, Loki-backed when available, with service, level and text filters. Falls back to Docker logs when Loki is not running. |
| Metrics | CPU, memory, request rate and latency percentiles (p50/p95/p99) from Prometheus, with a selectable time window. |
| Agent | Chat with an AI agent that has read access to logs, metrics and container state. Needs dashboard login or an ANTHROPIC_API_KEY. |
| Pipeline | Jenkins stage view and build history. Trigger a run, optionally skipping tests or the security scan. |
| Environments | ArgoCD sync state per environment, plus a live canary card: weight bar, step counter, Promote / Promote Full / Abort. |
| Deployments | Deploy history with git SHA, image tag, strategy, duration and p95 latency delta with regression flagging. |
| Plugins | Status of the project’s platform services. |
| Perf | Gatling load-test runs and their results. |
| Watcher | File-watching and rebuild activity. |
| Settings | Alert thresholds, log retention and metrics storage. |
Beyond the tabs, the header opens a Graph view of the system topology (tenant infrastructure, project lanes, services and the cluster components ArgoCD, Gitea and Argo Rollouts when a project runs on Kubernetes), plus a Terminal into the dashboard container.
The canary card
Section titled “The canary card”For --runtime kubernetes projects, the Environments tab renders the live Argo Rollout: current traffic split, which step it is on and buttons to promote or abort. It is the same rollout canary drives from the CLI, so you can start in one and finish in the other.
Compose-runtime projects have no Rollout, so the card does not appear.
The Agent tab
Section titled “The Agent tab”blissful-infra dashboard loginA one-time browser flow that authenticates the Agent tab against your Claude account, so you do not need to manage an API key. If you would rather use a key, set ANTHROPIC_API_KEY in your environment before starting the dashboard.
MCP server
Section titled “MCP server”The same capabilities are available to Claude Desktop and Claude Code over the Model Context Protocol:
{ "mcpServers": { "blissful-infra": { "command": "blissful-infra", "args": ["mcp"] } }}Once connected you can ask things like “what is the health of my running services?”, “show me ERROR logs from orders”, or “why did the last deploy regress p95?”.
Stopping it
Section titled “Stopping it”blissful-infra dashboard downThe dashboard runs as a detached container, so closing your terminal does not stop it.