Every service, every version.
Every server.
Discover every running service, installed package, container, database, and user account across your fleet. AI-structured into categorized entries with automatic version detection — searchable, exportable, always current.
On this page
Overview
Most inventory tools give you a list of packages and call it a day. ManageLM goes further: 14 deterministic checks collect raw system data, then the agent's LLM structures it into categorized inventory items with versions extracted from each service's own output — not guessed from package names.
The result is a browsable, searchable view of your infrastructure that answers real questions: "Which servers still run PostgreSQL 13?" "Do any machines have privileged Docker containers?" "Which boxes are missing our monitoring agent?"
No skill required. Inventory scans run directly from the portal — no skill assignment, no special permissions on the agent, no prompting. One button, one report.
What is collected
| Check | What it inspects |
|---|---|
| System Info | OS, kernel, uptime, CPU count, memory, disk usage |
| Running Services | Active systemd units (Linux) or Windows Services |
| Enabled Services | Services that start at boot |
| Listening Ports | TCP listening sockets with associated processes |
| Installed Packages | rpm / dpkg on Linux, installed programs on Windows |
| Package Versions | Explicit version extraction for nginx, PostgreSQL, Redis, Docker, & more |
| Containers | Docker / Podman containers with image, status, and port bindings |
| Cron Jobs | System crontab and per-user jobs |
| Network Interfaces | All interfaces with IP addresses and MAC |
| Filesystems | Non-virtual mounted filesystems |
| Hardware Info | CPU model, total memory, physical disks |
| Web Servers | Running nginx, Apache, Caddy, HAProxy instances |
| Databases | PostgreSQL, MySQL, MongoDB, Redis, Valkey, Memcached, Elasticsearch |
| Login Users | Non-system accounts with shell and group membership |
Smart categories
Raw check output gets categorized into twelve semantic buckets by the agent's LLM. Instead of a flat package list, you see your servers organized the way you actually think about them:
- system
- web
- database
- container
- network
- storage
- security
- monitoring
- log
- user
- scheduler
A server running nginx + PostgreSQL + Docker appears under web, database, and container — each category with its specific version and status, pulled from the service itself, not from the package manager.
How it works
Trigger
Open an agent's detail panel on the Infrastructure page and click the inventory icon. Or let a schedule do it automatically — daily, weekly, or monthly.Collect
The portal sends 14 deterministic check commands (defined inreports/inventory.json) to the agent over WebSocket. All checks run inside a Landlock read-only sandbox — the agent cannot modify disk during an inventory.Structure
The raw output is handed to the agent's configured LLM, which produces categorized, versioned inventory entries. This is the only built-in report that uses the LLM — 1 to 2 calls per scan.Browse
Results appear in the inventory modal, grouped by category. A fleet-wide view is one click away, and the same data powers the natural-language search interface for Claude, ChatGPT, and the portal.
Use cases
Legacy cleanup
"Find every server still running PostgreSQL 13." One natural-language query, an answer in seconds — no manual survey, no spreadsheet.
Migration planning
Know exactly what's deployed where before you plan a migration. Versions, dependencies, running services — all in one place.
Audit preparation
Export a fleet-wide PDF on demand. Auditors get a complete infrastructure inventory without you touching a single server.
Drift detection
Compare inventories over time to find unauthorized installs, missing monitoring agents, or services that shouldn't be running.
Capacity planning
Hardware info + running services + load give you the data to consolidate, right-size, and plan your next hardware refresh.
Incident response
When something breaks, you need to know what version is actually running — not what was last deployed. Inventory gives you ground truth.
Exports & scheduling
The Inventory button at the top of the Infrastructure page downloads a fleet-wide PDF — every server, every category, every version, ready for auditors or management. Use the Schedules popover to send it automatically (daily / weekly / monthly) to your admin users.
Scheduled inventories run on the same cadence as the email schedule, so reports are always fresh. A yellow D, W, or M badge on each agent card shows its active scan schedule.
Searchable via natural language. Inventory data feeds the search_inventory MCP tool, so Claude can answer questions like "Which public servers are running Redis older than 7.0?" directly in your chat — no dashboards, no queries.
Know your infrastructure.
Run your first inventory scan in 30 seconds. No setup, no skills, no agents to install beyond the ManageLM one.