PRTG Network Monitor Claude Code Skill

Prerequisites: Install the CLI

This skill drives the prtg-cli binary. You must verify the CLI is installed before invoking any command from this skill. If it is missing, install it first:

macOS / Linux:

bash <(curl -fsSL https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/prtg/install.sh)

Windows (PowerShell):

iwr -useb https://raw.githubusercontent.com/Servosity/msp-skills/main/skills/prtg/install.ps1 | iex

Verify: prtg-cli --version
Ensure ~/.local/bin (macOS / Linux) or %LOCALAPPDATA%\Programs\msp-skills (Windows) is on $PATH.

The installer downloads the prtg-cli and prtg-mcp binaries into ~/.local/bin (macOS / Linux) or %LOCALAPPDATA%\Programs\msp-skills (Windows). It does not register the skill with your agent and writes no MCP client config - see mcp-install.md for that wire-up.

If --version reports "command not found" after install, the runtime cannot see the binary directory on $PATH. Do not proceed with skill commands until verification succeeds.

Register each client's PRTG core once with 'core add', then run 'fleet alarms', 'fleet status', 'fleet handoff' or 'fleet drift' to see every core at once, and 'fleet ack', 'fleet pause' and 'fleet resume' to act across them with a preview first. Every v2 operation is also a typed command you can point at one core with --core.

When to Use This CLI

Use this CLI when an agent needs PRTG monitoring data or actions, especially across several PRTG cores (one per MSP client) registered with 'core add': current alarms, shift handoffs, maintenance pauses, acknowledgements and estate hygiene. It also exposes every PRTG API v2 operation for a single core. Fleet commands exit 0 when some cores fail (those rows carry an error) and non-zero only when every core fails or a sent action fails; read each row's error, action and failed fields. ack, pause and resume only preview until --yes is added.

Anti-triggers

Do not use this CLI for:

Do not use it for PRTG clusters or PRTG Hosted Monitor installations without API v2 (Paessler documents API v2 as unsupported on clusters); use the PRTG web interface or the legacy v1 HTTP API.
Do not use it to edit notification triggers, read the event log or tickets; API v2 does not expose them yet.
Do not use it to configure the PRTG server itself (installs, upgrades, core settings).

Known limitations

fleet alarms, fleet handoff, fleet silenced and fleet drift read PRTG's typed list endpoints (/sensors/alarms, /sensors, /devices, /groups, /probes). Paessler marks them deprecated in favour of /experimental/* but still serves them with no removal version; they also work on cores older than 24.3. sync reads the same lists.
The fleet commands were verified against mock cores, not a live PRTG installation.
PRTG API v2 is not available on PRTG clusters; Hosted Monitor support varies by version.

Unique Capabilities

One view across every core

fleet alarms - See every down, warning and unusual sensor across all of your clients' PRTG cores in one list: worst state first (down, warning, unusual, acknowledged), then higher priority, then oldest.

Reach for this at the start of a shift or when asked what is broken across clients.
```
prtg-cli fleet alarms --status down --unacked --agent
```
fleet status - One row per client core with reachability and sensor counts by state, worst first.

Use it for a quick health glance across every client before drilling into one.
```
prtg-cli fleet status --agent
```
fleet handoff - What changed across all cores since a point in time: new alarms, recoveries, and new acknowledgements or pauses with who and why.

Use it for shift handoffs and 'what happened overnight' questions.
```
prtg-cli fleet handoff --since 12h --agent
```
fleet silenced - Everything someone paused or acknowledged across all cores, with who, when, until when and why, flagging indefinite ones left too long.

Use it after maintenance windows and in weekly hygiene reviews to catch muted outages.
```
prtg-cli fleet silenced --indefinite --older-than 7d --agent
```
fleet exec - Run any read-only API v2 query on every core and get one merged table with a core column.

Use it for cross-client questions the dedicated fleet commands do not cover.
```
prtg-cli fleet exec /experimental/probes --agent
```

Estate hygiene

fleet drift - PRTG estate hygiene across cores: versions behind your newest core, expiring licenses or maintenance, sensor overuse, disconnected probes, rejected API keys and unreachable servers.

Use it for the weekly estate review instead of a hand-kept spreadsheet.
```
prtg-cli fleet drift --within 30d --agent
```
fleet sync - Refresh the offline mirror of every registered core in one run, each into its own isolated store.

Run it before offline searches or analytics across clients, or nightly from cron.
```
prtg-cli fleet sync --resources sensors,devices --agent
```

Safe actions across cores

fleet ack - Acknowledge matching down sensors on every core at once with a ticket message; previews per core and sends only with --yes (the API key needs the Acknowledge access level).

Use it to silence a known incident that spans several clients without opening each core.
```
prtg-cli fleet ack --filter 'name = "Ping"' --message 'TKT-4411 ISP outage' --duration 2h
```
fleet pause - Pause matching devices, groups, sensors or probes on every core for a maintenance window; previews per core and sends only with --yes (Write access level).

Use it before multi-client patch nights instead of pausing in every core's web UI.
```
prtg-cli fleet pause --kind device --filter 'name = "fw01"' --duration 2h --message 'Patch night'
```
fleet resume - Resume matching objects on every core after maintenance and optionally trigger an immediate scan; previews per core and sends only with --yes (Write access level).

Use it to close a maintenance window across clients and confirm everything comes back green.
```
prtg-cli fleet resume --kind device --filter 'name = "fw01"' --scan
```

Command Reference

ai - Manage ai

prtg-cli ai aistatus - Returns whether the PRTG AI Agent service is registered with the Windows Service Control Manager, whether it is running
prtg-cli ai analyze-aiincidents - Analyzes the currently active alarms with AI and returns potential incidents. This endpoint is experimental.

autodiscoveries - Manage autodiscoveries

prtg-cli autodiscoveries - Returns a list of all running auto-discoveries.

autodiscoverytasks - Manage autodiscoverytasks

prtg-cli autodiscoverytasks delete-auto-discovery-task - Deletes a given auto-discovery task This endpoint is experimental.
prtg-cli autodiscoverytasks get-auto-discovery-task - Returns auto-discovery task by the given task ID This endpoint is experimental.
prtg-cli autodiscoverytasks list-auto-discovery-task-overview - Returns a overview list of all auto-discovery tasks. This endpoint is experimental.
prtg-cli autodiscoverytasks modify-auto-discovery-task - Patch a auto-discovery task This endpoint is experimental.
prtg-cli autodiscoverytasks start-auto-discovery-task - Starts a auto-discovery task by the given task ID This endpoint is experimental.

channels - Manage channels

prtg-cli channels channels - Returns a list of channels. This endpoint is experimental.
prtg-cli channels get - Returns the metrics and settings of a channel.
prtg-cli channels list - Returns a list of channels. ⚠ Deprecated as of PRTG 24.3.
prtg-cli channels list-measurements - Get a list of the last measurements with references to the corresponding channels. ⚠ Deprecated as of PRTG 24.3.

core - Register PRTG cores for --core and the fleet commands

prtg-cli core add <name> --server <host[:port]> [--tag a,b] [--insecure-tls] - Register (or update) one core; its key goes in PRTG_FLEET_KEYS as <name>=<key>, never on disk; --insecure-tls skips certificate checks for that core only.
prtg-cli core list - List registered cores and whether each key env var is set (key_set).
prtg-cli core remove <name> - Remove a core and delete its local mirror and cache.

devices - Manage devices

prtg-cli devices create-sensor - Creates a new sensor in a device. Specify a sensor kind like 'ping', 'http' or 'paessler.snmp.traffic_sensor'.
prtg-cli devices delete - Deletes a device with all child sensors and channels. This endpoint is experimental.
prtg-cli devices get - Returns the metrics and settings of a device.
prtg-cli devices get-icons - Returns a list of device icons.
prtg-cli devices get-templates-experimental - Returns a list of device templates. This endpoint is experimental.
prtg-cli devices list - Returns a list of devices. ⚠ Deprecated as of PRTG 24.3.
prtg-cli devices list-experimental - Returns a list of devices. This endpoint is experimental.
prtg-cli devices metascan - Execute a meta-scan on a given device for a given sensor kind Some sensors use a metascan to retrieve important
prtg-cli devices metascan-result - Retrieve the meta scan progress/result This endpoint does not yet support all sensor kinds.
prtg-cli devices multi-pause - Pauses all devices that match a filter.
prtg-cli devices multi-resume - Resumes all devices that match a filter.
prtg-cli devices multi-scan-now - Triggers a scan of all devices that match a filter.
prtg-cli devices trigger-autodiscovery - Trigger autodiscovery for a given device This endpoint is experimental.
prtg-cli devices update - Edits a device.

groups - Manage groups

prtg-cli groups create-device-for - Creates a new device in a group. This endpoint is experimental.
prtg-cli groups create-for - Creates a new group in a group. This endpoint is experimental.
prtg-cli groups delete - Deletes a group with all child groups and devices. This endpoint is experimental.
prtg-cli groups get - Returns the metrics and settings of a group.
prtg-cli groups list - Get a list of all groups that match a filter. ⚠ Deprecated as of PRTG 24.3.
prtg-cli groups list-experimental - Returns a list of groups. This endpoint is experimental.
prtg-cli groups multi-pause - Pauses all groups that match a filter.
prtg-cli groups multi-resume - Resumes all groups that match a filter.
prtg-cli groups multi-scan-now - Triggers a scan of all groups that match a filter.
prtg-cli groups trigger-autodiscovery - Trigger autodiscovery for a given group This endpoint is experimental.
prtg-cli groups update - Edits a group.

health - Manage health

prtg-cli health - Returns the status of the PRTG application server.

libraries - Manage libraries

prtg-cli libraries - Returns a list of all libraries that match a filter. ⚠ Deprecated as of PRTG 24.3.

lookup-definitions - Manage lookup definitions

prtg-cli lookup-definitions get-overview - Returns a lookup definition. ⚠ Deprecated as of PRTG 24.3.100: This endpoint still works, but will be replaced.
prtg-cli lookup-definitions list - Returns a list of lookups definitions. ⚠ Deprecated as of PRTG 24.3.

objects - Manage objects

prtg-cli objects count - Returns a list of how many objects of each kind exist.
prtg-cli objects list - Get a list of all probes, groups, devices, and sensors that match a filter. ⚠ Deprecated as of PRTG 24.3.
prtg-cli objects multi-delete - Delete multiple objects based on the specified filter. All children will also be deleted. This endpoint is experimental.
prtg-cli objects objects - Returns a list of all probes, groups, devices, sensors, and channels that match a filter.

probe-network - Probe network information

prtg-cli probe-network <id> - Returns the network information of a probe.

probes - Manage probes

prtg-cli probes create-auto-discovery-task - Creates a new auto-discovery task on a probe This endpoint is experimental.
prtg-cli probes create-device-for - Creates a new device on a probe. This endpoint is experimental.
prtg-cli probes create-group-for - Creates a new group in a probe. This endpoint is experimental.
prtg-cli probes delete - Deletes a probe with all child groups and devices. This endpoint is experimental.
prtg-cli probes get - Returns the metrics and settings of a probe.
prtg-cli probes list - Get a list of all probes that match a filter. ⚠ Deprecated as of PRTG 24.3.
prtg-cli probes list-experimental - Returns a list of probes. This endpoint is experimental.
prtg-cli probes multi-pause - Pauses all probes that match a filter.
prtg-cli probes multi-resume - Resumes all probes that match a filter.
prtg-cli probes multi-scan-now - Triggers a scan of all probes that match a filter.
prtg-cli probes update - Edits a probe.

schemas - Manage schemas

prtg-cli schemas get - Returns a OAS schema to use in other endpoints. ⚠ Deprecated as of PRTG 24.3.
prtg-cli schemas get-create-by-parent-id - Returns an OAS schema for use in POST endpoints.
prtg-cli schemas get-kinds-for-parent - Returns a list of all object kinds that you can create on a given parent object. This endpoint is experimental.
prtg-cli schemas get-read-by-object-id - Returns OAS schema for use in GET endpoints for a specific object.
prtg-cli schemas get-update-by-object-id - Returns an OAS schema for use in PATCH endpoints for a specific object.

sensor-status-summary - Manage sensor status summary

prtg-cli sensor-status-summary global-sensor-summary - Returns a summary of all sensor states.
prtg-cli sensor-status-summary sensor-summary - Returns a summary of sensor states for a probe, group, or device.

sensors - Manage sensors

prtg-cli sensors delete - Deletes a sensor with all child channels. This endpoint is experimental.
prtg-cli sensors get - Returns metrics and settings of a sensor.
prtg-cli sensors list - Returns a list of all sensors that match a filter. ⚠ Deprecated as of PRTG 24.3.
prtg-cli sensors list-alarms - Returns a list of sensors with alarms that match a filter. ⚠ Deprecated as of PRTG 24.3.
prtg-cli sensors list-experimental - Returns a list of sensors. This endpoint is experimental.
prtg-cli sensors multi-acknowledge - Acknowledges the alarms of all sensors that match a filter.
prtg-cli sensors multi-pause - Pauses all sensors that match a filter.
prtg-cli sensors multi-resume - Resumes all sensors that match a filter.
prtg-cli sensors multi-scan-now - Triggers a scan of all sensors that match a filter.
prtg-cli sensors update - Edits a sensor. This endpoint does not yet support all sensor settings.

session - Manage session

prtg-cli session login - Creates a new user session.
prtg-cli session logout - Ends an active user session.
prtg-cli session renew - Renews an active user session.

setting-lookups - Manage setting lookups

prtg-cli setting-lookups <name> - Returns possible setting lookup values. ⚠ Deprecated as of PRTG 24.3.

settings - Manage settings

prtg-cli settings - Returns a list of settings that are available for unauthenticated users.

system - Manage system

prtg-cli system license-info - Returns PRTG license information.
prtg-cli system prtgfeature-flags - Returns a list of all feature-flags.
prtg-cli system prtgfeature-toggles - Returns a list of all enabled feature-toggles. ⚠ Deprecated as of PRTG 24.3.
prtg-cli system version-info - Returns information on the PRTG version and appserver version.

timeseries - Manage timeseries

prtg-cli timeseries graph-set - Returns time series data of a sensor for predefined time frames. This endpoint is experimental.
prtg-cli timeseries time-series - Get time series data ⚠ Deprecated as of PRTG 24.3.100: This endpoint still works, but will be removed.

usergroups - Manage usergroups

prtg-cli usergroups get-user-group - Returns the settings of a usergroup.
prtg-cli usergroups get-user-group-overview - Return information about a user group. ⚠ Deprecated as of PRTG 24.3.
prtg-cli usergroups list-user-groups - Get a list of all user groups that match a filter. ⚠ Deprecated as of PRTG 24.3.
prtg-cli usergroups list-user-groups-osk - Returns a list of user groups. This endpoint is experimental. See Overview: Maturity of endpoints (PRTG API v2 documentation)

users - Manage users

prtg-cli users delete-apikey - Deletes an API key.
prtg-cli users find-apikey - Returns details of an API key.
prtg-cli users get - Returns the metrics and settings of a user.
prtg-cli users get-overview - Returns information about a user account. ⚠ Deprecated as of PRTG 24.3.
prtg-cli users list - Returns a list of all user accounts. ⚠ Deprecated as of PRTG 24.3.100: This endpoint still works, but will be replaced.
prtg-cli users list-osk - Returns a list of users. This endpoint is experimental. See Overview: Maturity of endpoints (PRTG API v2 documentation)
prtg-cli users modify-apikey - Edits an API key.
prtg-cli users multi-pause-request - Pauses all user accounts that match a filter.
prtg-cli users multi-resume - Resumes all user accounts that match a filter.
prtg-cli users reset-password - Set a new password for a user account with the token from the password reset process.
prtg-cli users trigger-password-reset - Triggers the password reset process for a user account. Note: This route is not available when using PPHM.

fleet

prtg-cli fleet alarms - See every down, warning and unusual sensor across all of your clients' PRTG cores in one list: worst state first (down, warning, unusual, acknowledged), then higher priority, then oldest.
prtg-cli fleet status - One row per client core with reachability and sensor counts by state, worst first.
prtg-cli fleet handoff - What changed across all cores since a point in time: new alarms, recoveries, and new acknowledgements or pauses with who and why.
prtg-cli fleet silenced - Everything someone paused or acknowledged across all cores, with who, when, until when and why, flagging indefinite ones left too long.
prtg-cli fleet drift - PRTG estate hygiene across cores: versions behind your newest core, expiring licenses or maintenance, sensor overuse, disconnected probes, rejected API keys and unreachable servers.
prtg-cli fleet ack - Acknowledge matching down sensors on every core at once with a ticket message; previews per core and sends only with --yes (the API key needs the Acknowledge access level).
prtg-cli fleet pause - Pause matching devices, groups, sensors or probes on every core for a maintenance window; previews per core and sends only with --yes (Write access level).
prtg-cli fleet resume - Resume matching objects on every core after maintenance and optionally trigger an immediate scan; previews per core and sends only with --yes (Write access level).
prtg-cli fleet sync - Refresh the offline mirror of every registered core in one run, each into its own isolated store.
prtg-cli fleet exec - Run any read-only API v2 query on every core and get one merged table with a core column.

Finding the right command

When you know what you want to do but not which command does it, ask the CLI directly:

prtg-cli which "pause monitoring for patch night"

which resolves a natural-language capability query to the best matching command from this CLI's curated feature index. Exit code 0 means at least one match; exit code 2 means no confident match - fall back to --help or use a narrower query. --json (and other machine formats) keep that exit-2 contract and write {"matches":[]} on stdout so agents can inspect the envelope without treating a miss as success.

Recipes

Shift-start sweep

prtg-cli fleet alarms --status down,warning --unacked --agent

Unacknowledged down and warning sensors across every client, highest priority and oldest first.

Narrow a large sensor list

prtg-cli sensors list-experimental --core acme --filter 'status = down' --agent --select id,name,status,status_since

Keep only the fields an agent needs from one core's sensor list.

Overnight handoff

prtg-cli fleet handoff --since 12h --agent

New alarms, recoveries and new acknowledgements or pauses across cores in the last 12 hours.

Forgotten pauses

prtg-cli fleet silenced --indefinite --older-than 7d --agent

Objects paused or acknowledged with no end date for more than a week, with who did it.

Weekly estate review

prtg-cli fleet drift --within 30d --agent

Version skew, licenses and maintenance ending within 30 days, sensor overuse and disconnected probes.

Auth Setup

Each PRTG core has its own API key (PRTG: Account Settings > My Account > API Keys), sent as 'Authorization: Bearer '. For one core, set PRTG_SERVER to its host[:port] (the PRTG application server listens on 1616 by default) and PRTG_API_KEY to the key. Fleet commands only see cores registered with 'core add --server <host[:port]>'; put every core's key in PRTG_FLEET_KEYS as name=key pairs separated by commas (acme=KEY1,globex=KEY2), and keys are never written to disk. Check registered cores with 'core list' (key_set must be true) and 'doctor --core '. A registered core with a self-signed certificate takes 'core add --insecure-tls' (that core only); for single-core use pass --insecure or PRTG_SKIP_TLS_VERIFY=true.

Run prtg-cli doctor (single core), or prtg-cli core list plus prtg-cli doctor --core <name> for registered cores. Fleet commands only read cores added with core add; with none registered they return an empty list.

Agent Mode

Add --agent to any command. Expands to: --json --compact --no-input --no-color.

Global format flags share one contract on promoted, novel, sync, and --deliver paths:

--json - one JSON document on stdout (sync progress events go to stderr)
--compact - keep identity/status/timestamp fields; does not change the document vs stream shape
--csv / --plain - tabular rows (collection envelopes unwrap to the row array). --csv prefixes ' on a cell or header key that starts with =, +, -, @, tab, CR, or LF (fullwidth ＝＋－＠ too) so a spreadsheet does not run it as a formula; a plain signed number such as -5 or +2.5e3 is unchanged.
--quiet - one identity value per row, no envelope
Pipeable - JSON on stdout, errors on stderr
Filterable - --select keeps a subset of fields. Dotted paths descend into nested structures; arrays traverse element-wise. Critical for keeping context small on verbose APIs:
```
prtg-cli autodiscoveries --agent --select active,autodiscovery_type,last_autodiscovery
```
Previewable - --dry-run shows the request without sending
Offline-friendly - sync/search commands can use the local SQLite store when available
Non-interactive - never prompts, every input is a flag
Explicit confirmation - --agent does not imply --yes; pass --yes separately only after the target, arguments, and side effects are clear
Explicit retries - use --idempotent only when an already-existing create should count as success, and use --ignore-missing only when a missing delete target should count as success

Response envelope

Commands that read from the local store or the API wrap output in a provenance envelope:

{
  "meta": {"source": "live" | "local", "synced_at": "...", "reason": "..."},
  "results": <data>
}

Parse .results for data and .meta.source to know whether it's live or local. A human-readable N results (live) summary is printed to stderr only when stdout is a terminal AND no machine-format flag (--json, --csv, --compact, --quiet, --plain, --select) is set - piped/agent consumers and explicit-format runs get pure JSON on stdout.

Paths and state

Agents should treat the CLI's path resolver as part of the runtime contract:

Use --home <dir> for one invocation, or set PRTG_HOME=<dir> to relocate all four path kinds under one root.
Use per-kind env vars only when a specific kind must diverge: PRTG_CONFIG_DIR, PRTG_DATA_DIR, PRTG_STATE_DIR, PRTG_CACHE_DIR.
Resolution order is per-kind env var, --home, PRTG_HOME, XDG (XDG_CONFIG_HOME, XDG_DATA_HOME, XDG_STATE_HOME, XDG_CACHE_HOME), then platform defaults.
config contains settings like config.toml and profiles. data contains credentials.toml, data.db, and auth sidecars. config also holds fleet.json (registered cores) and per-core homes under cores/<name>/. state contains persisted queries, jobs, and teach.log. cache contains regenerable HTTP/cache files.
Stored secrets live in credentials.toml under the data dir. Existing legacy config.toml secrets are read for compatibility and leave config.toml on the first auth write.
Run prtg-cli doctor --fail-on warn to surface path and credential-location warnings. agent-context exposes a schema v4 paths block for agents that need the resolved dirs.

For MCP, pass relocation through the MCP host config. The MCP binary does not inherit CLI flags:

{
  "mcpServers": {
    "prtg": {
      "command": "prtg-mcp",
      "env": {
        "PRTG_HOME": "/srv/prtg"
      }
    }
  }
}

Precedence across machines: an inherited per-kind env var overrides an explicit --home for that kind (with --core, --home and --config are rejected and each core uses its own home). Use PRTG_HOME or per-kind vars as durable levers, and use --home only for a single invocation. Relocation is not reversible by unsetting env vars; move files manually before clearing PRTG_HOME, or doctor will not find credentials left under the former root.

Local data is unencrypted and relies on OS account permissions plus disk encryption. Reads in the default --data-source auto mode are written to the SQLite store; GET responses are cached for five minutes under the cache dir. local-data purge previews what it would erase (store files, response cache, learn journal) and deletes only with --yes; it is not exposed over MCP. PRTG_NO_LOCAL_DATA=1 disables the response cache and store writes, turns auto reads into live reads, and blocks sync and workflow archive; --data-source local still reads an existing store, and PRTG_DATA_SOURCE sets the --data-source default.

Automatic learning

This CLI ships a self-capturing learning loop. The CLI does its own bookkeeping: every invocation is journaled locally, a failed flag followed by a corrected retry auto-derives a flag_alias candidate, and a teach on a query family without a playbook auto-synthesizes a playbook_candidate from the session's journal. Your job is judgment only: recall first, act on surfaced candidates, teach the final answer, playbook amend when you observe a correction. You never record failures by hand.

Step 1: `recall` before any discovery

Before list/search/drill commands on a new user question, pass the question as an argv or MCP tool argument to recall --agent. Do not interpolate user-controlled text into a shell command line.

Quoted recall "<question>" breaks on an apostrophe, which is ordinary English. A quoted heredoc breaks when a body line equals the delimiter, and that delimiter is published in these docs. Write the question with a non-shell file-writing tool, then read it back as data:

# Write the question verbatim with your file-writing tool (no shell involved).
# Command substitution on a file only ever yields data  -  the shell never
# parses the file's bytes as syntax.
QUERY=$(cat /path/to/question.txt)
prtg-cli recall "$QUERY" --agent

Prefer MCP: pass the question as the tool's query argument. "$QUERY" after a file read is argv-safe; putting the question itself in the command text is not.

The response envelope:

{
  "query": "...",
  "normalized": "<normalized form>",
  "query_entities": ["..."],
  "found": true | false,
  "match_score": 0.0,
  "results": [
    { "resource_id": "...", "resource_type": "...", "venue": "...",
      "confidence": 2, "entity_match": "exact|partial|unknown",
      "source": "taught|preseed|pattern", "warnings": ["..."] }
  ],
  "mismatches": [ /* only when --debug-mismatches */ ],
  "warnings": [ /* top-level */ ],
  "candidates": [
    { "id": 12, "class": "flag_alias | playbook_candidate",
      "summary": "...", "sightings": 3, "last_seen": "...",
      "rationale": "...",
      "next_action": ["<trial command>", "prtg-cli learnings confirm 12"] }
  ],
  "playbook": {
    "query_family": "...",
    "playbook": {
      "steps": [ { "cmd": "<command with {slot} substitution>", "purpose": "..." } ],
      "entity_slots": ["$ENTITY"],
      "expected_tool_calls": 3
    },
    "slots_resolved": { "$ENTITY": { "token": "<live token>", "canonical": "<canonical>" } },
    "notes": "<workarounds + gotchas for this query family>"
  },
  "notes": "<duplicate surface for non-playbook callers>"
}

Empty-store short-circuit: if the store has no learnings, playbooks, or candidates yet (recall finds nothing and learnings list and learnings candidates are both empty), skip recall for the rest of this session instead of taxing every query; resume recall-first once something has been taught.

Step 2: decision tree

Read candidates, playbook, notes, results[0], and warnings in that order:

if Candidates present (warnings include "candidates_present"):
    -> candidates are try-then-confirm, never facts. Follow each candidate's
       two-step next_action verbatim: run the trial command first, then run
       `learnings confirm <id>` only after the trial verified the behavior.
       Reject a wrong candidate with `learnings reject <id>`.
    -> NEVER re-teach something recall surfaced as a candidate; confirm or
       reject that candidate instead of teaching a duplicate.
    -> candidates ride alongside playbooks and resource hits, not instead of
       them; continue with the branches below after acting on them.

if Playbook present:
    -> READ Playbook.notes verbatim FIRST (workarounds + gotchas the CLI surface doesn't expose)
    -> replay Playbook.steps in order, substituting Playbook.slots_resolved entries
       for the entity slot tokens. If a step's slot is unresolved, fall back to
       discovery for that step only.
    -> the Playbook's expected_tool_calls is a budget; if you find yourself running
       materially more, record the divergence via `prtg-cli playbook amend`
       at end-of-session.

elif Notes present (no Playbook):
    -> read Notes verbatim before any discovery step; they carry known gotchas
       for this query family even when no structured choreography exists yet.

elif Found AND Results[0].EntityMatch == "exact" AND Results[0].Confidence >= 2:
    -> skip discovery; fetch live data for Results[*].ResourceID in parallel

elif Found AND Results[0].EntityMatch == "partial":
    -> candidate hint, NOT a hit; read the resource title to validate before trusting

elif (any row in Mismatches[] when --debug-mismatches was passed):
    -> treat as cold start; the stored learning is for a different entity
       (different canonical resolved from query_entities)

else:  // Found == false, no playbook, no notes
    -> cold start; run discovery normally; teach the answer afterward (Step 4).
       If the family has no playbook yet, that teach auto-synthesizes a
       playbook candidate from this session's journal - you do not need to
       record one by hand.

Playbook and Notes are orthogonal to the per-resource path. A recall response can carry both a Playbook AND a Results[] hit - use both: the Playbook tells you which choreography to run; the resource hits short-circuit specific steps. Default to skipping mismatches; pass --debug-mismatches only when investigating cold-start surprises.

Candidate judgment details: learnings confirm <id> prints the candidate's full payload before materializing it - check that the printed payload matches the behavior you verified. learnings reject <id> tombstones the derivation signature so the same candidate does not resurface. The envelope carries only the few candidates worth acting on now; prtg-cli learnings candidates lists the full open set.

Graceful degradation: if learnings confirm is an unknown command, you are driving an older binary - ignore the candidates guidance and follow the rest of the protocol.

Step 3: always read `warnings`

low_confidence: row exists at confidence<2. Treat as a hint, not a skip-discovery hit.
resource_not_in_store: the local store doesn't have the resource the learning points at. The match validator couldn't classify entities - direct-fetch and re-evaluate.
cross_alias_match (per-result): the row was taught under a different alias and matched the live query's canonical via entity_lookups (e.g., an "acked" teach satisfying a "down acknowledged" recall). Trust the resource_id.
similar_shape_different_entity:<canonical> (top-level): a structurally matching row exists but its canonical entity differs from the live query's. Treated as cold start; the warning carries the conflicting canonical as a hint, but the row is NOT promoted into Results.
ambiguous_alias (top-level): a single query entity resolved to multiple canonicals (e.g., a client alias that maps to two registered cores). Surface the ambiguity from context before committing to a resource.
candidates_present (top-level): the envelope carries a candidates section. Handle it via the candidates branch in Step 2 before anything else.
lookup_refresh_available (top-level): an entity in the query has no lookup row yet, but synced data could provide one. Run prtg-cli sync to refresh entity lookups.
Top-level no_learnings_for_query_family: the table had no rows above the Jaccard floor. Pure cold start.

Step 4: `teach &` after finalizing your response - always

Teaching is unconditional. After resolving a query the store could not answer, background-teach the final resource mapping - no call-count threshold, no judging whether it was "worth" learning. The teach is the anchor of the loop: it triggers playbook synthesis for a family without a playbook, and same-referent phrasings fold into one family so near-duplicate teaches do not fragment the store. Fire it after assembling your user-facing response but BEFORE emitting it, with a shell & so the call returns immediately. Pass the query the same way as recall - argv/MCP, or file-then-$QUERY. Do not splice the question into the command text:

QUERY=$(cat /path/to/question.txt)
prtg-cli teach --query "$QUERY" --resource-type sensor --resource 2045 --resource 2046
# (append shell `&` to background it)

Silent on success. Errors only land in teach.log under the resolved state dir. Teach the most specific resource - if the user asked a broad question and you walked through parent records to find the specific answer, teach the leaf id, not the parent. The CLI uses seeded entity_lookups for cross-alias resolution at recall time, so a teach under one alias (e.g., "Niners") satisfies future queries under another alias (e.g., "49ers", "San Francisco") automatically.

PII rule: teach the structural question with identifiers stripped - never include names, emails, phone numbers, account ids, or other personal identifiers in taught queries or notes. The CLI scans teach queries for obvious email/phone shapes and warns, but does not block; strip before teaching rather than relying on the warning.

Step 5: playbooks - optional flags, automatic synthesis

You do not need to decide whether a session "deserves" a playbook: a teach on a family without one auto-synthesizes a playbook_candidate from the session's journal, and the next session judges it via confirm/reject. Attach explicit playbook flags only when you already hold choreography worth recording verbatim - workarounds the CLI didn't surface (silently-dropped flags, undocumented params, pagination tricks, payload gotchas). Prefer the integrated one-call form - record the resource learning and the playbook in the same teach invocation:

# Common case: record both the resource learning AND the playbook in one call.
QUERY=$(cat /path/to/question.txt)
prtg-cli teach \
  --query "$QUERY" \
  --resource <id> \
  --playbook-file ~/playbooks/fleet-alarms.json \
  --playbook-notes-file ~/playbooks/fleet-alarms-notes.md
# (append shell `&` to background it)

# Alternate: playbook-only (no resource to record alongside).
QUERY=$(cat /path/to/question.txt)
prtg-cli teach-playbook \
  --query "$QUERY" \
  --playbook-file ~/playbooks/fleet-alarms.json \
  --notes-file ~/playbooks/fleet-alarms-notes.md

Playbook files are JSON with steps, entity_slots, expected_tool_calls. Notes files are markdown carrying the gotchas verbatim. File-free callers (MCP-only agents) pass the same content inline: --playbook-json and --playbook-notes on the integrated teach form, --playbook-json and --notes on teach-playbook. On the integrated teach form, the playbook flags are optional - omit them entirely for a resource-only teach. On the standalone teach-playbook form, at least one of the playbook and notes flags must be set; both empty is rejected. Playbooks are keyed on the structural query family (entities stripped) so a recipe taught from one entity-shaped query applies to every other query of the same shape, with slots_resolved binding the live query's canonical at recall time.

When you DO find a playbook on a future recall, treat it as ground truth: replay the steps with slots_resolved substitutions, skip the discovery that the choreography already documents, and read notes before any step.

Step 6: `playbook amend &` when your debug response identifies a correction

If your debug-protocol response identifies a concrete correction the notes or playbook should know - a workaround, an undocumented endpoint shape, a stale field name, observed schema drift, an empty-payload fallback - fire playbook amend BEFORE emitting your user-facing response. Same fire-and-forget posture as teach. Pass the query and note as argv/MCP arguments, or write each with a non-shell file tool and read them back (QUERY=$(cat ...), NOTE=$(cat ...)). Do not interpolate either string into the command text:

QUERY=$(cat /path/to/question.txt)
NOTE=$(cat /path/to/note.txt)
prtg-cli playbook amend \
  --query "$QUERY" \
  --add-note "$NOTE"
# (append shell `&` to background it)

What counts as worth amending: a behavior you OBSERVED this session that future-you would benefit from knowing. Examples worth amending:

A workaround for a CLI surface that silently drops or misorders a flag.
An undocumented endpoint shape (response wrapped in {meta, results}, payload nested two levels deeper than the docs claim).
Observed schema drift (a field renamed, an index that shifted between PRTG versions, a category label that the API now returns lower-cased).

What does NOT belong in notes:

The year-specific or entity-specific answer to the user's question. That's the response, not a learning.
Per-core / per-sensor / per-row data the playbook already retrieves at runtime.
Statements that paraphrase what the existing notes already say.

The amend command appends to the family's existing notes with a timestamped marker ([amend YYYY-MM-DDTHH:MMZ]: <text>). Multiple amends accumulate; the audit trail is visible. If no playbook exists yet for the family, amend creates a notes-only one (so cold-start corrections still land).

PII discipline for amend notes

playbook amend notes are designed to potentially flow upstream as shared knowledge in future versions of the Printing Press. Keep them clean of user-identifying content so the upstream-contribution path stays open without retroactive scrubbing:

Do NOT embed paths to user filesystems, personal API keys or tokens, user email addresses, user GitHub handles, or specific query histories tied to a single user.
Acceptable: endpoint shapes, undocumented field names, API gotchas, observed schema drift, workarounds for CLI surfaces, generalizable pagination or retry tactics.

If a correction is only meaningful with user-specific context, it belongs in a personal note, not in the playbook amend.

Measuring the loop

prtg-cli learnings stats reports recall hit rate, teach-to-reuse, playbook resolution rate, and candidate confirm/reject counts from the local learn_events table. Rates are null until they have a denominator; everything stays on this machine. Use it to check whether the loop is earning its keep for this CLI.

Disabling learning

--no-learn on a single command short-circuits both recall and the teach write path. Use for deterministic agent flows or tests that must not be affected by accumulated learnings.
PRTG_NO_LEARN=true in the environment globally disables the pipeline.

Agent Feedback

When you (or the agent) notice something off about this CLI, record it:

prtg-cli feedback "the --since flag is inclusive but docs say exclusive"
prtg-cli feedback --stdin < notes.txt
prtg-cli feedback list --json --limit 10

Entries are stored locally as feedback.jsonl under the resolved data dir. They are never POSTed unless PRTG_FEEDBACK_ENDPOINT is set AND either --send is passed or PRTG_FEEDBACK_AUTO_SEND=true. Default behavior is local-only.

Write what surprised you, not a bug report. Short, specific, one line: that is the part that compounds.

Output Delivery

Every command accepts --deliver <sink>. The output goes to the named sink in addition to (or instead of) stdout, so agents can route command results without hand-piping. Three sinks are supported:

Sink	Effect
`stdout`	Default; write to stdout only
`file:<path>`	Atomically write output to `<path>` as a private 0600 file (exclusive randomly named temp in the destination directory, then rename; a planted symlink is not followed). Binary-response commands write decoded payload bytes (not the base64 JSON envelope) and print a small JSON receipt on stdout; `--json`/`--csv` do not refuse when this sink is set.
`webhook:<url>`	POST the output body to the URL (`application/json`). Redirects are not followed; a 3xx response fails the delivery and the body is not sent to the `Location`.

Unknown schemes are refused with a structured error naming the supported set. Webhook failures, including redirects, return non-zero and log the URL and HTTP status on stderr.

Named Profiles

A profile is a saved set of flag values, reused across invocations. Use it when a scheduled or recurring agent reuses the same saved flags while providing different input each run.

prtg-cli profile save briefing --json
prtg-cli --profile briefing autodiscoveries
prtg-cli profile list --json
prtg-cli profile show briefing
prtg-cli profile delete briefing --yes

Explicit flags always win over profile values; profile values win over defaults. agent-context lists all available profiles under available_profiles so introspecting agents discover them at runtime.

Async Jobs

For endpoints that submit long-running work, the generator detects the submit-then-poll pattern (a job_id/task_id/operation_id field in the response plus a sibling status endpoint) and wires up three extra flags on the submitting command:

Flag	Purpose
`--wait`	Block until the job reaches a terminal status instead of returning the job ID immediately
`--wait-timeout`	Maximum wait duration (default 10m, 0 means no timeout)
`--wait-interval`	Initial poll interval (default 2s; grows with exponential backoff up to 30s)

Use async submission without --wait when you want to fire-and-forget; use --wait when you want one command to return the finished artifact.

If --wait times out or polling fails after the job was accepted, the command exits 8 and prints the job ID plus a recovery command that fetches the result. The job may still be running and may already be billed, so run the recovery command instead of resubmitting.

Exit Codes

Code	Meaning
0	Success
2	Usage error (wrong arguments)
3	Resource not found
4	Authentication required
5	API error (upstream issue)
6	Partial failure
7	Rate limited (wait and retry)
8	Async job submitted but not finished (run the printed recovery command; do not resubmit)
10	Config error

Argument Parsing

Parse $ARGUMENTS:

Empty, help, or --help → show prtg-cli --help output
Starts with install → ends with mcp → MCP installation; otherwise → see Prerequisites above
Anything else → Direct Use (execute as CLI command with --agent)

MCP Server Installation

Install the MCP binary (run the install script from the Prerequisites section, or see mcp-install.md for per-agent wire-up).

claude mcp add prtg-mcp --env PRTG_SERVER=prtg.acme.example:1616 --env PRTG_API_KEY="$PRTG_API_KEY" -- prtg-mcp

Add --env PRTG_SKIP_TLS_VERIFY=true for a core with a self-signed certificate.

claude mcp list

Verify: claude mcp list

For a remote host, prtg-mcp --transport http requires PRTG_MCP_HTTP_TOKEN set to at least 32 characters (callers send Authorization: Bearer <token>; openssl rand -hex 32 prints 64 hex characters and meets that minimum), serves http://127.0.0.1:7777/mcp by default, and needs --tls-cert/--tls-key for any non-loopback --addr. Requests with no Origin header are accepted. Loopback origins are accepted; any other Origin is refused unless listed in PRTG_MCP_HTTP_ALLOWED_ORIGINS (comma-separated scheme://host[:port] values). PRTG_MCP_READ_ONLY=1 drops mirrored write commands (local-write commands that only touch the CLI's own learn store stay) and makes the raw execute tool refuse state-changing calls.

Direct Use

Check if installed: which prtg-cli If not found, offer to install (see Prerequisites at the top of this skill).
Match the user query to the best command from the Unique Capabilities and Command Reference above.

Execute with the --agent flag:

prtg-cli fleet alarms --status down --agent

If ambiguous, drill into subcommand help: prtg-cli fleet --help.

Ask in your favorite AI

Documentation