# Full Infrastructure Discovery

> Full self-hosted or enterprise infrastructure discovery pipeline. Scans an AgentConfig-managed inventory via tunnel-manager, discovers containers via configured runtime providers, maps networking and DNS, and ingests a privacy-safe topology snapshot into the Knowledge Graph. Use for a governed fleet topology refresh; do not use for ad hoc host probing.

- Skill: `knuckles-team/full-infrastructure-discovery` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add knuckles-team/full-infrastructure-discovery`
- Raw SKILL.md: https://api.skillmd.com/api/skills/knuckles-team/full-infrastructure-discovery/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Product & Planning
- Author: Knuckles-Team (https://skillmd.com/u/knuckles-team)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/knuckles-team/full-infrastructure-discovery

---


# full-infrastructure-discovery Workflow

Discover a self-hosted or enterprise deployment from AgentConfig connection
references and ingest a privacy-safe topology snapshot. Environment topology stays
in the configured providers rather than this workflow.

### Step 0: discover-hosts [skill: tunnel-manager-mcp]
List all hosts from inventory and check SSH connectivity
Expected: host, inventory, connectivity

### Step 1: container-manager-mcp
List all containers across all discovered hosts with their images, status, ports, and networks
Expected: container, image, status
Depends On: Step 0

### Step 2: portainer-mcp
List all Portainer environments and stacks with their deployment status
Expected: environment, stack, endpoint

### Step 3: scan-host-networks [skill: tunnel-manager-mcp]
Run network interface scan across all hosts to discover subnets, VLANs, and VPN tunnels
Expected: network, interface, subnet
Depends On: Step 0

### Step 4: technitium-dns-mcp
List all DNS records to map domain-to-service routing
Expected: record, domain

### Step 5: graph-os
Ingest discovered topology into the Knowledge Graph as HardwareNode, Container, ContainerStack, NetworkSubnet, DNSRecord nodes with RUNS_ON, BELONGS_TO_STACK, RESOLVES_DNS_FOR relationships
Expected: ingest, node, relationship
Depends On: Step 1, Step 2, Step 3, Step 4

## Execution

Run this workflow as a dependency-ordered DAG. Steps with no unmet `depends_on` run in parallel; dependents run after their prerequisites complete.

- **Run first (in parallel):** Step 0 — discover-hosts; Step 1 — container-manager-mcp; Step 2 — portainer-mcp; Step 3 — scan-host-networks; Step 4 — technitium-dns-mcp; Step 5 — graph-os

**Execution:** If graph-os is reachable, offload the whole DAG via `graph_orchestrate action=execute_workflow` (or the `kg-delegate` skill) for true parallel/swarm execution. Otherwise execute the steps natively in dependency order: run steps with no unmet `depends_on` in parallel, then their dependents.