This server provides read-only MCP tools to monitor on-premises Linux VMs and PostgreSQL databases over SSH.
VM & Service Management
list_vms — Discover all configured VMs with reachability status and watched service names
get_service_status — Check systemd state, uptime, and last 5 log lines for a service
get_service_logs — Retrieve bounded journald logs filtered by severity level and line count
get_log_file — Fetch the last N lines of an allowed log file, with optional grep filtering
get_infra_overview — Combined snapshot of service statuses and database health for a VM in one call
Database Inspection
get_db_status — View connection counts, waiting locks, and long-running queries for a PostgreSQL database
query_db — Execute caller-supplied SELECT statements (read-only, row-limited)
list_tables — List all tables (schema + name) in a database, with optional cache refresh
describe_table — Inspect a table's columns, types, primary key, and foreign keys
Auditing
get_audit_log — View recent entries from the local append-only audit log of all remote operations
Safety & Security
All DB queries run in a READ ONLY transaction; SSH commands are checked against per-VM allowlists
Log file access is restricted to an allowed directory list (path traversal blocked)
All output is bounded server-side (200 log lines, 100 DB rows, 200 tables max) with truncation flagged
Provides read-only access to PostgreSQL databases over SSH, allowing agents to query tables, describe schemas, list tables, and check database health metrics such as connection counts and locks.
Read-only MCP access to on-prem Linux VMs and PostgreSQL databases over SSH.
agent ──stdio──▶ infra-mcp ──SSH──▶ VMs (journald · log files)
└──▶ DBs (read-only PostgreSQL)An agent can check service health, retrieve bounded logs, inspect DB state, and explore table schemas — without terminal access. Every remote operation is allowlist-gated and written to an append-only audit log.
uv tool install infra-mcp
# or
pip install infra-mcpRelated MCP server: mcp-infra-readonly
Copy infra-mcp.yaml.example to ~/.infra-mcp/infra-mcp.yaml and edit it.
# Generate a starter config from ~/.ssh/config
infra-mcp generate-config -o ~/.infra-mcp/infra-mcp.yaml
# Create the read-only PostgreSQL role(s)
infra-mcp setup
# Check VM reachability
infra-mcp test
# Refresh discovered services, log dirs, and databases (updates config in place)
infra-mcp discover --in-placeOverride the config path with --config or INFRA_MCP_CONFIG.
infra-mcp runRegister as a stdio MCP server in your client (Claude Code, Cursor, …) with command infra-mcp run.
infra-mcp checks PyPI for a newer release once a day and, when one exists, prints a
one-line hint to stderr telling you how to upgrade:
uv tool upgrade infra-mcp # or: pip install --upgrade infra-mcpThe check runs in the background, never blocks startup, and never touches stdout. Print
the installed version with infra-mcp --version. Disable the check entirely by setting
INFRA_MCP_NO_UPDATE_CHECK=1.
Tool | Purpose |
| All VMs with reachability and watched services |
| Service states + DB health for one VM in a single call |
| systemd state, uptime, last 5 log lines |
| Bounded journald logs, filtered by severity |
| Last N lines of an allowed log file, optional grep |
Tool | Purpose |
| Connection counts, waiting locks, long-running query count |
| Bounded caller-supplied |
| Tables in a database (schema + name), capped at 200 |
| Columns, types, primary key, foreign keys for one table |
Tool | Purpose |
| Recent entries from the local audit log |
All output is bounded server-side (200 log lines, 100 DB rows, 200 tables/columns max). Truncation is always flagged with a -- TRUNCATED: marker. list_tables and describe_table cache results in memory (TTL: schema_cache_ttl_hours, default 24 h); pass refresh: true to force a live re-read.
SSH commands and systemd services are checked against a per-VM allowlist before any network call.
All DB queries run as a read-only role inside a READ ONLY transaction.
Log file paths are resolved against a per-VM directory allowlist (.. traversal blocked).
Every remote operation is appended to a local JSONL audit log.
Describes one named table — its columns and types, primary key, and foreign-key relationships — scoped to just that table, so you can author a correct (possibly multi-table) query directly. Indexes and comments are not included. Set refresh=true to force a live re-read.
Inputs: db (required), table (required), refresh. Returns: result
Returns recent entries from the local audit log (oldest first).
Inputs: lines. Returns: result
Returns health metrics for a configured database: connection counts, waiting locks, and long-running query count.
Inputs: db (required). Returns: result
Returns service statuses and database health for a named VM in a single call. Call this first when diagnosing any VM-level issue. Unreachable services or databases are listed with an error note rather than failing the whole call.
Inputs: vm (required). Returns: result
Returns the last N lines of a log file on a VM, with optional grep pattern. Both grep and line-capping execute on the VM before any data is transmitted.
Inputs: vm (required), path (required), lines, pattern. Returns: result
Returns bounded journald logs for a service, filtered by severity level. Filtering and line-capping execute on the VM before transmission.
Inputs: vm (required), service (required), level, lines. Returns: result
Returns systemd state, uptime, and last 5 log lines for a service on a VM. Use for a quick health check before requesting full logs.
Inputs: vm (required), service (required). Returns: result
Lists the tables of a configured database with their schema/namespace, in one bounded call — no column detail. Call this first to decide which tables are relevant before describing them. Set refresh=true to force a live re-read.
Inputs: db (required), refresh. Returns: result
Returns all configured VMs with reachability status and watched service names. Call this first in any session to discover available infrastructure.
Inputs: no named parameters. Returns: result
Executes a caller-supplied SELECT statement against a configured database, bounded by row limit. Only SELECT is permitted.
Inputs: db (required), sql (required), limit. Returns: result
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 10 tools
Each tool targets a distinct resource or action: tables (list, describe, query), VMs (list, overview), services (status, logs), logs (file, service), database health, audit log. No overlap in purpose.
All tool names follow a consistent verb_noun pattern (describe_table, get_audit_log, list_vms, etc.), making it easy to infer functionality.
10 tools is appropriate for infrastructure diagnostics, covering VMs, databases, services, logs, and audit without being excessive or sparse.
The tool set provides complete coverage for read-only infrastructure diagnosis: discovering VMs, checking health, exploring database schemas, querying data, and retrieving logs. No obvious gaps for its diagnostic purpose.
Read-only MCP access to a documented IT fleet: state, changes, posture. 15 tools.
Hosted MCP server for PostgreSQL diagnostics: slow queries, missing indexes, connection pressure.
Linux server administration over OAuth MCP: shell, Docker, systemd, files and monitoring.
Read-only server insight: why a Linux server or K8s cluster is slow, failing or changed.
