Skip to main content
Glama
Pexip Infinity MCP Server
by Josh-E-S

CI License: MIT Python 3.11+ Ruff

MCP (Model Context Protocol) server for the Pexip Infinity Management API. It gives an LLM-based agent everything it needs to manage and operate a Pexip Infinity deployment, exposing 122 curated tools and covering all four admin API categories: Configuration, Status, History, and Command.

Disclaimer: This is an independent, community-built project. It is not affiliated with, endorsed by, or sponsored by Pexip. It uses Pexip's public Management API and sends no telemetry or analytics. The server needs no external network access beyond the Pexip Management Node you configure, plus your own identity provider if you choose to use OIDC. It never sends your data anywhere else.

Quick start

You'll need your Pexip Management Node hostname plus credentials. OAuth2 is recommended for production; basic auth (admin username + password) is recommended only for dev/lab environments. See Configuration to set up an OAuth2 client.

The server starts read-only by default: it can list and report, but every create / update / delete / control tool is removed from the catalog until you explicitly enable writes. Connect first, confirm it works, then decide.

① Claude Desktop: one-click install

The simplest path: no terminal, no JSON.

  1. Download the .mcpb bundle for your OS (macOS or Windows) from the latest release.

  2. Double-click it. Claude Desktop opens an install dialog showing the server and the permissions it requests.

  3. Fill in the form with your Management Node host and either an OAuth2 client (recommended) or an admin username + password (dev/lab only). Leave read-only on. Click install.

  4. Try it. Ask Claude: "List the VMRs on my Pexip node" or "Show me the Pexip system status."

To enable writes later, edit the server in Claude Desktop's settings and turn read-only off.

② Manual config: Claude Code & other MCP hosts

For Claude Code, Cursor, or any MCP host that takes a JSON server config.

1. Install the server (needs Python 3.11+):

git clone https://github.com/Josh-E-S/pexip-mgmt-mcp.git
cd pexip-mgmt-mcp
pipx install .          # installs the `pexip-mgmt-mcp` command on your PATH

2. Add it to your MCP host by pointing its config at the pexip-mgmt-mcp command. Use OAuth2 for production:

{
  "mcpServers": {
    "pexip-mgmt": {
      "command": "pexip-mgmt-mcp",
      "env": {
        "PEXIP_HOST": "pexip-mgr.example.com",
        "PEXIP_AUTH_MODE": "oauth2",
        "PEXIP_OAUTH2_CLIENT_ID": "your-client-id",
        "PEXIP_OAUTH2_PRIVATE_KEY": "-----BEGIN PRIVATE KEY-----\n...\n-----END PRIVATE KEY-----\n"
      }
    }
  }
}

For dev/lab, use basic auth instead: swap the PEXIP_AUTH_MODE / PEXIP_OAUTH2_* keys for "PEXIP_USERNAME": "admin" and "PEXIP_PASSWORD": "...". See Configuration for all options.

  • Enable writes: add "PEXIP_READ_ONLY": "false" to env (read-only is the default).

  • Claude Code shortcut (instead of hand-editing JSON):

    # OAuth2 (recommended for production)
    claude mcp add pexip-mgmt \
      -e PEXIP_HOST=pexip-mgr.example.com \
      -e PEXIP_AUTH_MODE=oauth2 \
      -e PEXIP_OAUTH2_CLIENT_ID=your-client-id \
      -e PEXIP_OAUTH2_PRIVATE_KEY="$(cat oauth2_private_key.pem)" \
      -- pexip-mgmt-mcp
    
    # basic auth (dev/lab only)
    claude mcp add pexip-mgmt \
      -e PEXIP_HOST=pexip-mgr.example.com \
      -e PEXIP_USERNAME=admin \
      -e PEXIP_PASSWORD=... \
      -- pexip-mgmt-mcp

Other ways to run

  • Docker (self-hosted HTTP transport): run the container alongside your Infinity, serving on 127.0.0.1:8000. For lab/dev, fronting it with a tunnel or proxy is fine. For production, run it as a proper service behind a hardened reverse proxy (TLS, authentication, access control) on a managed host, not a dev tunnel. See DEPLOY.md.

  • From source (developers): clone the repo, then:

    pip install -e .
    cp .env.example .env          # edit with your Management Node details
    python -m pexip_mcp --healthcheck
    # OK: connected to pexip-mgr.example.com as admin, schema fetched

    See Testing for the full dev workflow.

Related MCP server: Equinix Fabric MCP

Contents

Coverage

API

Tools

What it covers

Configuration

46

VMRs + aliases, end users, devices, gateway rules, automatic participants, IVR themes, LDAP sync, locations + Conferencing Nodes, global settings, schema introspection, plus generic CRUD (list/get/create/update/delete_resource) over ~70 registry resources: SIP/H.323/MS-SIP proxies, TURN/STUN, Teams Connectors, Azure tenants, Google Meet tokens, MJX (endpoints, integrations, deployments), DNS/NTP/SMTP/syslog/SNMP, certificates (CA/TLS/CSR), admin roles + identity providers, backups + upgrades, web app hosting, policy profiles, and more

Status

39

Live conferences + per-node shards, participants + media streams, per-participant call quality, registrations, node/location status + load stats, backplanes, alarms, licensing, cloud overflow, Exchange scheduler, MJX endpoints + meetings, Teams Connector nodes + calls

History

14

Conference + participant CDRs, server-side aggregation (summarize_calls), alarm history, backplane history, registration history, node event history

Command

23

Active call control: dial, disconnect, mute/unmute (participant or all guests), set role, lock/unlock, transfer, layout, messages, LDAP sync, provisioning emails, backup create/restore, certificate import, snapshot, upgrade

The full catalog with per-tool annotations and parameters is in TOOLS.md (regenerate with uv run python scripts/generate_tools_md.py).

This MCP server is designed to optimize tool use. High-traffic resources (VMRs, users, devices, rules) get dedicated typed tools; the ~70 remaining configuration resources share five generic CRUD tools backed by a resource registry. That keeps the catalog at 122 tools, which lowers per-request token use and improves the agent's tool selection (verified by the eval suite).

How it fits together

Source layout:

src/pexip_mcp/
├── client.py                 # Async PexipClient (all 4 API categories), 429-aware retry
├── config.py                 # PexipSettings (env-driven, pydantic-settings)
├── mcp_app.py                # FastMCP instance + lifespan
├── server.py                 # Imports tool modules to trigger registration
├── __main__.py               # Entry point + --healthcheck + --http
└── tools/
    ├── _helpers.py           # get_client, resolve_id_by_field, paginate_all, annotation presets
    ├── command.py            # Active call control + name→UUID resolvers
    ├── status.py             # Live state + per-participant quality
    ├── history.py            # CDRs + summarize_calls aggregation
    ├── resource_crud.py      # Generic CRUD + resource registry (~70 resources)
    ├── schema.py             # Live schema introspection
    └── conference.py, end_user.py, device.py, gateway_rule.py, alias.py,
        automatic_participant.py, infrastructure.py, ldap.py,
        ivr_theme.py, global_settings.py   # dedicated typed CRUD per resource

Examples

Real Claude Desktop conversations against a lab Pexip node.

A status question, answered directly:

Config inventory across multiple resources in one ask:

CDRs turned into an interactive dashboard, not a data dump:

Configuration

All env vars are PEXIP_* prefixed and loaded from .env (or the process environment). See .env.example for an annotated template.

Variable

Default

Purpose

PEXIP_HOST

required

Management Node hostname or IP, no scheme

PEXIP_AUTH_MODE

basic

oauth2 (recommended for production) or basic (dev/lab)

PEXIP_OAUTH2_CLIENT_ID

oauth2 only

OAuth2 Client ID from the Pexip UI

PEXIP_OAUTH2_PRIVATE_KEY

oauth2 only

OAuth2 client "Private key" (ES256, shown once)

PEXIP_OAUTH2_TOKEN_URL

https://<host>/oauth/token/

Override the token endpoint

PEXIP_OAUTH2_SCOPE

is_admin use_api

OAuth2 scopes requested

PEXIP_USERNAME

basic only

Admin username (typically admin)

PEXIP_PASSWORD

basic only

Admin password

PEXIP_VERIFY_TLS

true

Set false for self-signed lab nodes

PEXIP_TIMEOUT

30

HTTP timeout in seconds

PEXIP_MAX_RETRIES

3

Retries on 429 rate-limit responses

PEXIP_READ_ONLY

true

Expose only read tools; remove all write/control tools. On by default; set false to enable writes

PEXIP_ALLOW_SECURITY_RESOURCES

false

Allow generic CRUD to mutate security-critical resources (SSH keys, roles, auth, certs). Only relevant when writes are enabled

PEXIP_ALLOW_PLATFORM_TOOLS

false

Expose platform-lifecycle command tools (backup/restore, upgrade, cert import, software upload, cloud-node start, snapshot). Removed at startup unless enabled, even with writes on

PEXIP_MCP_AUTH_MODE

token

How HTTP clients authenticate to this server: token (static bearer) or oauth (OIDC). Only affects --http. See docs/identity.md

PEXIP_MCP_TOKEN

unset

Bearer token for the --http transport (required for non-loopback binds; min 32 chars). Run pexip-mgmt-mcp --generate-token

PEXIP_OIDC_ISSUER

unset

Issuer URL; required when PEXIP_MCP_AUTH_MODE=oauth (validate OIDC JWTs from your own IdP: Entra/Google/Okta/on-prem)

PEXIP_OIDC_AUDIENCE

unset

Expected token audience; required when PEXIP_MCP_AUTH_MODE=oauth

PEXIP_OIDC_REQUIRED_SCOPES

unset

Space-separated scopes the JWT must carry (optional)

PEXIP_OIDC_JWKS_URI

discovered

Override the JWKS endpoint (default: discovered from the issuer)

Read-only mode (default)

The server runs in read-only mode by default: only the read tools (list / get / schema) are exposed. Every create, update, delete, and Command-API control tool is removed from the catalog at startup, so the LLM cannot mutate the deployment even if it tries. This is enforced server-side: the tools are gone from the catalog, not merely flagged with an advisory readOnlyHint.

To enable the mutating admin surface, set PEXIP_READ_ONLY=false (logged loudly at startup). Even then, generic CRUD refuses to touch security-critical resources (SSH keys, admin roles/permissions, authentication/SSO, TLS/CA certs) unless you also set PEXIP_ALLOW_SECURITY_RESOURCES=true. Pair writes with a least-privilege Pexip Administrator Role for defense in depth.

Authentication: OAuth2 vs basic

Two modes, selected by PEXIP_AUTH_MODE:

  • oauth2 (recommended for production): an OAuth2 JWT bearer assertion (ES256). The server signs a short-lived JWT with the client's private key and exchanges it at https://<host>/oauth/token/ for a 1-hour bearer token (cached and auto-refreshed). It authenticates once per hour instead of on every request and keeps a reusable admin password out of the server's environment.

  • basic (dev/lab only): a local Management Node admin username + password. Works out of the box on every Infinity deployment and is the quickest way to get started, but sends admin credentials on each request. Not recommended for production.

Set up an OAuth2 client in the Pexip admin UI under Users & Devices > OAuth2 Clients: add a client, then copy its Client ID and Private key (the private key is shown only once). You also enable Management API OAuth2 and attach an Administrator Role, per Pexip's Managing API access via OAuth2. Then set PEXIP_AUTH_MODE=oauth2, PEXIP_OAUTH2_CLIENT_ID, and PEXIP_OAUTH2_PRIVATE_KEY.

Quality: the eval suite

The evals/ suite measures whether an LLM can drive these tools correctly from natural language. It defines 156 scenarios written as real admin requests ("mute all the guests in AllHands", "point the syslog server at 10.0.0.99"), graded automatically across three layers:

Layer

What runs

Cost

Deterministic (332 checks)

Every eval case is validated against the live tool registry (tools exist, parameter names match signatures)

free, in CI

LLM-graded (--llm)

Each prompt goes to Claude with the full tool catalog; multi-turn conversations with mocked API responses; graded on tool choice, parameters, and chain order

~$2 / full run (tool catalog is prompt-cached)

Live (--live, 33 tests)

CRUD, status reads, and call commands against a real Infinity node, including auto-dialing a test call and moderating it by name

free (your lab)

Scoring supports exact / subset / ordered-subset / any-of tool matching, per-step parameter checks with acceptable-alternative values, and optional steps for chains where server-side name resolution makes a lookup legitimate but unnecessary. See evals/README.md for the case format and how to add scenarios.

Current state: the full suite passes: 698 passed, 12 skipped, 0 failures across all four layers (208 unit + 332 deterministic eval checks + LLM-graded + 33 live), 86% coverage. The 12 skips are env-gated live cases (e.g. no dial target set). Reproduce the full run with uv run pytest tests/ evals/ --llm --live, or the free deterministic subset (540 checks) with uv run pytest tests/ evals/.

Skills SDK

pexip-mgmt-skills/ is a self-contained Agent Skills (open standard) + Claude Code plugin package that wraps this MCP server with operator runbooks and developer-reference skills. Built to be extractable: cp -r pexip-mgmt-skills/ out and you have a complete, plug-installable SDK that loads in Claude Code, Gemini CLI, Codex CLI, Cursor, or any other compliant host.

Currently ships 9 skills across 6 domains:

Domain

Skill

Audience

router

pexip-mgmt-intake

both, start here for open-ended requests

operations

pexip-operations

operator: kick / lock / report / configure

management-api

pexip-config-api

developer: modify Configuration API tool code

management-api

pexip-status-api

developer: Status API

management-api

pexip-history-api

developer: History API

management-api

pexip-command-api

developer: Command API

events

pexip-event-sinks

both, configure Pexip's webhook push-event destinations

policy

pexip-external-policy

developer: external policy server config (via generic CRUD)

room-integration

pexip-mjx

both, One-Touch Join

See pexip-mgmt-skills/README.md for the install instructions and pexip-mgmt-skills/ARCHITECTURE.md for the design rules. The companion awesome-pexip-skills covers the client-side (webapp3, @pexip/infinity, @pexip/media). Install both for full Pexip coverage.

Testing

uv run pytest                  # 208 unit + 332 deterministic eval checks, 86% coverage
uv run pytest tests/           # 208 unit tests (mocked HTTP via respx), ~2s
uv run pytest evals/           # 332 deterministic eval checks, free
uv run pytest evals/ --llm     # LLM-graded evals (needs ANTHROPIC_API_KEY, ~$2)
uv run pytest evals/ --live    # integration against a real node (needs .env)
uv run ruff check src tests evals

Every run prints a coverage report, but the 80% gate is enforced only in CI on the full suite (--cov-fail-under=80 in .github/workflows/ci.yml), so subset runs like uv run pytest evals/ show partial coverage without failing. Unit tests mock the Pexip Management API with respx; the retry suite monkeypatches asyncio.sleep so backoff tests run instantly.

License

MIT

Available Tools

76 tools
get_active_participantA
Read-onlyIdempotent

Retrieve a single active participant.

Args: participant_id: UUID, or the participant's display name (e.g. "Bob") — names are resolved against currently connected participants automatically, so there is no need to call list_active_participants first. conference: Optional conference name to scope a display-name lookup when the name might not be unique across meetings.

Inputs: participant_id (required), conference. Returns: object

get_alarm_historyA
Read-onlyIdempotent

Retrieve one historical alarm by id.

Inputs: alarm_id (required). Returns: object

get_backplaneA
Read-onlyIdempotent

Retrieve a single backplane by id.

Inputs: backplane_id (required). Returns: object

get_backplane_historyA
Read-onlyIdempotent

Retrieve one historical backplane connection by id.

Inputs: backplane_id (required). Returns: object

get_backplane_history_media_streamsA
Read-onlyIdempotent

Get media stream statistics for a historical backplane connection.

Args: backplane_id: Backplane id from list_backplane_history.

Inputs: backplane_id (required). Returns: object

get_backplane_media_streamsA
Read-onlyIdempotent

Get media stream statistics for a backplane connection.

Args: backplane_id: Backplane id from list_backplanes.

Inputs: backplane_id (required). Returns: object

get_cloud_monitored_locationB
Read-onlyIdempotent

Retrieve a single cloud monitored location by id.

Inputs: location_id (required). Returns: object

get_cloud_nodeA
Read-onlyIdempotent

Retrieve a single cloud overflow node by id.

Inputs: node_id (required). Returns: object

get_cloud_overflow_locationA
Read-onlyIdempotent

Retrieve a single cloud overflow location by id.

Inputs: location_id (required). Returns: object

get_conference_shardB
Read-onlyIdempotent

Retrieve a single conference shard by id.

Inputs: shard_id (required). Returns: object

get_conference_sync_statusA
Read-onlyIdempotent

Retrieve a single conference sync entry by id.

Inputs: sync_id (required). Returns: object

get_conferencing_nodeA
Read-onlyIdempotent

Retrieve a Conferencing Node by integer id or by exact name.

Inputs: node (required). Returns: object

get_deviceA
Read-onlyIdempotent

Retrieve a device by integer id or by exact alias.

Inputs: device (required). Returns: object

get_end_userA
Read-onlyIdempotent

Retrieve an end user by integer id or by primary_email_address.

Inputs: user (required). Returns: object

get_exchange_scheduler_statusA
Read-onlyIdempotent

Retrieve a single Exchange scheduler entry by id.

Inputs: scheduler_id (required). Returns: object

get_gateway_ruleA
Read-onlyIdempotent

Retrieve a gateway routing rule by integer id or exact name.

Inputs: rule (required). Returns: object

get_global_settingsA
Read-onlyIdempotent

Retrieve the platform-wide global configuration singleton.

Includes things like default themes, default join PIN behavior, bandwidth caps, banner text, MSSIP domain, and call-create permissions.

Inputs: no named parameters. Returns: object

get_history_conferenceB
Read-onlyIdempotent

Retrieve one past conference instance by id.

Inputs: conference_id (required). Returns: object

get_history_participantA
Read-onlyIdempotent

Retrieve one past participant by id.

Returns full quality forensics: historic_call_quality (per-window timeline) and bucketed_call_quality ([unknown, good, ok, bad, terrible] counts). These deep fields are only populated on individual GET, not list responses.

Inputs: participant_id (required). Returns: object

get_ivr_themeA
Read-onlyIdempotent

Retrieve an IVR theme by id or exact name.

Inputs: theme (required). Returns: object

get_ldap_sourceA
Read-onlyIdempotent

Retrieve an ldap_sync_source by id or exact name.

Includes last sync status / errors when available — useful for answering 'is LDAP sync working?'.

Inputs: source (required). Returns: object

get_licensing_statusA
Read-onlyIdempotent

Current concurrent port usage vs entitlement (audio + video ports, per location).

Inputs: no named parameters. Returns: object

get_locationA
Read-onlyIdempotent

Retrieve a system location by integer id or by exact name.

Inputs: location (required). Returns: object

get_location_statisticsA
Read-onlyIdempotent

Get load statistics for one system location (call counts, port usage).

Args: location: Location id or exact name.

Inputs: location (required). Returns: object

get_location_statusA
Read-onlyIdempotent

Get live status for one system location by id or exact name.

Inputs: location (required). Returns: object

get_management_node_statusB
Read-onlyIdempotent

Retrieve live status for a specific Management Node.

Inputs: node_id (required). Returns: object

get_mjx_endpoint_statusA
Read-onlyIdempotent

Retrieve live status for a single MJX endpoint by id.

Inputs: endpoint_id (required). Returns: object

get_mjx_meeting_statusA
Read-onlyIdempotent

Retrieve status for a single MJX meeting by id.

Inputs: meeting_id (required). Returns: object

get_node_event_historyA
Read-onlyIdempotent

Retrieve one Conferencing Node status event by id.

Inputs: event_id (required). Returns: object

get_node_statisticsA
Read-onlyIdempotent

Get load statistics for one Conferencing Node (CPU, media load, call counts).

Args: node: Node id or exact name.

Inputs: node (required). Returns: object

get_node_statusA
Read-onlyIdempotent

Get live status for one Conferencing Node by integer id or exact name.

Inputs: node (required). Returns: object

get_participant_qualityA
Read-onlyIdempotent

Live call quality for one active participant.

Combines two Status API endpoints into one response:

  • the participant record (call_quality, connect_time, conference, role, protocol, remote_address, location, current packet loss summary)

  • all media streams for the participant (per-stream rx/tx bitrate, packet loss, jitter, codec, resolution).

For "is Alice's call OK right now?" use cases — pass the name directly. For post-call quality forensics, use get_history_participant which exposes bucketed_call_quality.

Args: participant_id: UUID, or the participant's display name (e.g. "Bob") — names are resolved against currently connected participants automatically, so there is no need to call list_active_participants first. conference: Optional conference name to scope a display-name lookup when the name might not be unique across meetings.

Inputs: participant_id (required), conference. Returns: object

get_registration_aliasA
Read-onlyIdempotent

Retrieve a single registered alias by id.

Inputs: alias_id (required). Returns: object

get_registration_historyA
Read-onlyIdempotent

Retrieve one historical registration entry by id.

Inputs: entry_id (required). Returns: object

get_resourceA
Read-onlyIdempotent

Get a single configuration resource by integer id or by name.

Args: resource: API resource name (see list_resources for the full list). id: Integer id, numeric string, or the resource's name — names are resolved automatically, so there is no need to list first.

Inputs: id (required), resource (required). Returns: object

get_resource_schemaA
Read-onlyIdempotent

Fetch the live JSON schema for a Pexip configuration resource.

Use this to discover available fields, types, required vs optional (nullable: false = required), allowed enum values, filtering options, and help text for any resource. Pass the resource name as it appears in the URL path, e.g. "conference", "conference_alias", "end_user", "system_location", "worker_vm", "gateway_routing_rule", "automatic_participant".

Inputs: resource (required). Returns: object

get_teams_node_call_statusA
Read-onlyIdempotent

Retrieve status for a single Teams Connector call by id.

Inputs: call_id (required). Returns: object

get_teams_node_statusA
Read-onlyIdempotent

Retrieve status for a single Teams Connector node by id.

Inputs: node_id (required). Returns: object

get_vmrA
Read-onlyIdempotent

Retrieve a VMR by integer id or by exact name.

Inputs: vmr (required). Returns: object

list_active_conferencesA
Read-onlyIdempotent

List currently-running conference instances.

Empty objects list = nothing is in progress right now.

Args: name: Filter by exact instance name (the dialed alias). service_type: "conference" / "lecture" / "two_stage_dialing" / "gateway" / "test_call". tag: Filter by service tag. limit: Per-page limit (ignored if fetch_all). offset: Pagination offset (ignored if fetch_all). fetch_all: Walk all pages and return up to 5,000 records combined.

Inputs: tag, name, limit, offset, +2 more. Returns: object

list_active_participantsB
Read-onlyIdempotent

List currently-connected participants (active call legs).

Args: conference_name: Restrict to one running conference instance. role: "chair" or "guest". protocol: "api" / "sip" / "h323" / "mssip" / "webrtc" / "rtmp" / "teams" / "gms". is_muted: Filter by mute state. limit, offset, fetch_all: Pagination controls.

Inputs: role, limit, offset, is_muted, +3 more. Returns: object

list_alarm_historyA
Read-onlyIdempotent

List historical alarms.

Args: start_time: Lower bound (inclusive) on time_raised. UTC ISO 8601. end_time: Upper bound (exclusive) on time_raised. UTC ISO 8601. level: "error" / "warning" / "info". limit, offset, fetch_all: Pagination controls.

Inputs: level, limit, offset, end_time, +2 more. Returns: object

list_alarmsA
Read-onlyIdempotent

List active platform alarms.

Args: level: "error" / "warning" / "info". node_name: Restrict to one Conferencing Node by name. limit, offset, fetch_all: Pagination controls.

Inputs: level, limit, offset, fetch_all, +1 more. Returns: object

list_aliasesA
Read-onlyIdempotent

List conference aliases, optionally scoped to one VMR.

Args: vmr: VMR id or exact name to filter by. Omit for all aliases. alias: Exact alias match. alias_contains: Case-insensitive substring match. limit: Max results. offset: Pagination offset.

Inputs: vmr, alias, limit, offset, +1 more. Returns: object

list_automatic_participantsA
Read-onlyIdempotent

List automatic participants, optionally scoped to one VMR.

Args: vmr: VMR id or name to filter by (pass the name directly — resolved internally, no need to list_vmrs first). Omit for all. alias_contains: Case-insensitive substring match on alias. limit: Max results. offset: Pagination offset.

Inputs: vmr, limit, offset, alias_contains. Returns: object

list_backplane_historyC
Read-onlyIdempotent

List historical backplane connections.

Inputs: limit, offset, fetch_all. Returns: object

list_backplanesB
Read-onlyIdempotent

List backplane status (inter-node media connections).

Inputs: limit, offset, fetch_all. Returns: object

list_cloud_monitored_locationsA
Read-onlyIdempotent

List all locations monitored for dynamic bursting.

Inputs: limit, offset, fetch_all. Returns: object

list_cloud_nodesB
Read-onlyIdempotent

List all cloud overflow Conferencing Nodes (dynamic bursting).

Inputs: limit, offset, fetch_all. Returns: object

list_cloud_overflow_locationsA
Read-onlyIdempotent

List all locations containing Conferencing Nodes for dynamic bursting.

Inputs: limit, offset, fetch_all. Returns: object

list_conference_shardsA
Read-onlyIdempotent

List conference instances broken out per Conferencing Node (shards).

Args: conference_name: Filter by conference name. limit, offset, fetch_all: Pagination controls.

Inputs: limit, offset, fetch_all, conference_name. Returns: object

list_conference_sync_statusD
Read-onlyIdempotent

List conference synchronization status.

Inputs: limit, offset, fetch_all. Returns: object

list_conferencing_nodesA
Read-onlyIdempotent

List Conferencing Nodes (the worker_vm resource).

Args: location: System location id or exact name to filter by. name_contains: Case-insensitive substring match on node name. node_type: "CONFERENCING" or "PROXYING". limit: Max results. offset: Pagination offset.

Inputs: limit, offset, location, node_type, +1 more. Returns: object

list_devicesA
Read-onlyIdempotent

List provisioned devices (registration records).

Args: alias_contains: Case-insensitive substring match on the device alias. owner_email: Filter by primary owner's email address (exact). tag: Filter by tag. limit: Max results. offset: Pagination offset.

Inputs: tag, limit, offset, owner_email, +1 more. Returns: object

list_end_usersA
Read-onlyIdempotent

List directory end users.

Args: email_contains: Case-insensitive substring match on primary_email_address. name_contains: Case-insensitive substring match on display_name. sync_tag: Filter by LDAP sync tag. limit: Max results. offset: Pagination offset.

Inputs: limit, offset, sync_tag, name_contains, +1 more. Returns: object

list_exchange_scheduler_statusC
Read-onlyIdempotent

List Secure Scheduler for Exchange integration status.

Inputs: limit, offset, fetch_all. Returns: object

list_gateway_rulesA
Read-onlyIdempotent

List dial plan gateway routing rules, ordered by priority ascending.

Args: name_contains: Case-insensitive substring match on rule name. enabled_only: If true, returns only enabled rules. limit: Max results. offset: Pagination offset.

Inputs: limit, offset, enabled_only, name_contains. Returns: object

list_history_conferencesA
Read-onlyIdempotent

List completed conference instances.

Args: start_time: Lower bound (inclusive) on conference start_time. ISO 8601 in UTC, e.g. "2026-05-07T00:00:00". end_time: Upper bound (exclusive) on conference start_time. UTC ISO 8601. name: Filter by exact instance name. service_type: "conference" / "lecture" / etc. tag: Filter by service tag. limit, offset, fetch_all: Pagination controls.

Inputs: tag, name, limit, offset, +4 more. Returns: object

list_history_participantsA
Read-onlyIdempotent

List completed participant call legs (CDR-like records).

Time bounds apply to participant.start_time. All times are UTC ISO 8601.

Args: start_time: Lower bound (inclusive) on start_time. end_time: Upper bound (exclusive) on start_time. conference_name: Restrict to one conference name (string, not id). call_direction: "in" / "out". call_quality: "1_good" / "2_ok" / "3_bad" / "4_terrible". protocol: "sip" / "h323" / "mssip" / "webrtc" / "rtmp" / "teams" / etc. disconnect_reason: e.g. "Call disconnected", "Call failed", "Call rejected". location: system_location id or exact name. service_tag: Filter by tag carried from the parent conference. limit, offset, fetch_all: Pagination. fetch_all caps at 10,000.

Inputs: limit, offset, end_time, location, +8 more. Returns: object

list_ivr_themesB
Read-onlyIdempotent

List IVR themes (branding bundles assignable to VMRs).

Inputs: limit, offset, name_contains. Returns: object

list_ldap_sourcesB
Read-onlyIdempotent

List configured LDAP / Active Directory sync sources.

Inputs: limit, offset, name_contains. Returns: object

list_locationsB
Read-onlyIdempotent

List system locations (logical groupings of Conferencing Nodes, typically per datacenter / region).

Inputs: limit, offset, name_contains. Returns: object

list_location_statusA
Read-onlyIdempotent

List live status for system locations.

Args: name_contains: Case-insensitive substring match on location name. limit, offset, fetch_all: Pagination controls.

Inputs: limit, offset, fetch_all, name_contains. Returns: object

list_management_node_statusC
Read-onlyIdempotent

List Management Node live status.

Inputs: limit, offset. Returns: object

list_mjx_endpoint_statusA
Read-onlyIdempotent

List One-Touch Join endpoint live status.

Args: name_contains: Case-insensitive substring match on endpoint name. limit, offset, fetch_all: Pagination controls.

Inputs: limit, offset, fetch_all, name_contains. Returns: object

list_mjx_meeting_statusB
Read-onlyIdempotent

List One-Touch Join meeting status (upcoming meetings detected by MJX).

Inputs: limit, offset, fetch_all. Returns: object

list_node_event_historyB
Read-onlyIdempotent

List Conferencing Node status event history.

Args: start_time: Lower bound (inclusive). UTC ISO 8601. end_time: Upper bound (exclusive). UTC ISO 8601. limit, offset, fetch_all: Pagination controls.

Inputs: limit, offset, end_time, fetch_all, +1 more. Returns: object

list_node_statusA
Read-onlyIdempotent

Live status (load, sync, version, boot time) for each Conferencing Node.

Args: location: Filter to one system_location by name or id. name_contains: Case-insensitive substring match on node name. limit, offset, fetch_all: Pagination controls.

Inputs: limit, offset, location, fetch_all, +1 more. Returns: object

list_registration_aliasesA
Read-onlyIdempotent

List currently registered aliases on the platform.

Args: alias_contains: Case-insensitive substring match on the alias. limit, offset, fetch_all: Pagination controls.

Inputs: limit, offset, fetch_all, alias_contains. Returns: object

list_registration_historyB
Read-onlyIdempotent

List historical registration events.

Args: alias_contains: Case-insensitive substring match on the alias. limit, offset, fetch_all: Pagination controls.

Inputs: limit, offset, fetch_all, alias_contains. Returns: object

list_registrationsA
Read-onlyIdempotent

List endpoints currently registered to the platform (live Status API read).

This is the runtime counterpart to the device records: who is registered right now, on what node and protocol — not the configured allow-list.

Args: alias_contains: Case-insensitive substring match on the registered alias. protocol: "sip" / "h323" / "mssip" / "webrtc". limit, offset, fetch_all: Pagination controls (fetch_all walks all pages).

Inputs: limit, offset, protocol, fetch_all, +1 more. Returns: object

list_resourcesA
Read-onlyIdempotent

List any Pexip configuration resource that has no dedicated tool.

This is the catch-all listing tool. Supported resources by area:

  • Admin auth & roles: authentication, role, ldap_role, permission, adfs_auth_server, adfs_auth_server_domain, identity_provider, identity_provider_group, user_group, user_group_entity_mapping.

  • Integrations & signaling: teams_proxy (Microsoft Teams Connectors), azure_tenant (Teams CVI tenant), gms_access_token, gms_gateway_token (Google Meet), ms_exchange_connector, exchange_domain, sip_proxy, sip_credential, mssip_proxy, h323_gatekeeper, turn_server, stun_server, policy_server, policy_profile, break_in_allow_list_address, telehealth_profile.

  • Scheduling & sync: registration, conference_sync_template, ldap_sync_field, scheduled_conference, scheduled_alias, recurring_conference.

  • Media library: media_library_entry, media_library_playlist, media_library_playlist_entry.

  • Platform & certificates: management_vm, licence, licence_request, media_processing_server, diagnostic_graphs, ca_certificate, tls_certificate, certificate_signing_request.

  • MJX / One-Touch Join: mjx_integration, mjx_endpoint, mjx_endpoint_group, mjx_meeting_processing_rule, mjx_exchange_deployment, mjx_exchange_autodiscover_url, mjx_graph_deployment, mjx_google_deployment.

  • System infrastructure: dns_server, ntp_server, http_proxy, syslog_server, snmp_network_management_system, smtp_server, static_route, ssh_authorized_key.

  • Backups & upgrades: system_backup (existing system backups), autobackup (backup schedule), upgrade, software_bundle, software_bundle_revision, scheduled_scaling.

  • Web app: webapp_alias, webapp_branding, external_webapp_host.

Call get_resource_schema(resource) to discover filterable fields.

Args: resource: API resource name from the list above (e.g. "sip_proxy", "system_backup", "teams_proxy", "dns_server"). name_contains: Case-insensitive substring match on the name field. filters: Additional query filters as key-value pairs (e.g. {"enable": true}). Keys must match the resource's field names. limit: Max results per page (default 20). offset: Pagination offset.

Inputs: resource (required), limit, offset, filters, +1 more. Returns: object

list_teams_node_call_statusC
Read-onlyIdempotent

List Teams Connector call status.

Inputs: limit, offset, fetch_all. Returns: object

list_teams_node_statusC
Read-onlyIdempotent

List Teams Connector node status.

Inputs: limit, offset, fetch_all. Returns: object

list_vmrsA
Read-onlyIdempotent

List VMRs (Pexip conference objects with service_type=conference).

Args: name: Exact name match. name_contains: Case-insensitive substring match on name. tag: Filter by tag. limit: Max results (default 20). Use 0 to return all (avoid on large platforms). offset: Pagination offset.

Inputs: tag, name, limit, offset, +1 more. Returns: object

summarize_callsA
Read-onlyIdempotent

Summarize / report on call history — grouped counts and totals over a window.

USE THIS for any "summarize", "report", "breakdown", "how many calls by X", "stats", or "totals" request. Returns aggregated counts + durations grouped by a field — NOT individual records. Prefer this over list_history_conferences / list_history_participants whenever the user wants aggregates rather than a raw list; it walks pagination internally and is far cheaper on context and tool calls than fetching individual records.

Args: start_time: Lower bound (inclusive) on start_time. UTC ISO 8601 required. end_time: Upper bound (exclusive) on start_time. UTC ISO 8601 required. group_by: One of "call_direction", "call_quality", "protocol", "service_tag", "system_location", "conference_name", "disconnect_reason", "vendor". conference_name: Restrict to one conference name. service_tag: Restrict to one tag. call_direction: "in" or "out" — useful when group_by is something else. location: system_location filter (name or id). max_records: Hard cap on participants fetched. Defaults to 10,000 (the platform retention limit).

Returns: { "total_calls": int, "total_duration_seconds": int, "average_duration_seconds": float, "time_range": {"start": ..., "end": ...}, "group_by": str, "groups": {: {"count": int, "duration_seconds": int}, ...}, "truncated": bool # true if max_records was hit before exhausting results }

Inputs: end_time (required), start_time (required), group_by, location, +4 more. Returns: object

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    B3.2/5.0

    Scored across 76 tools

    Disambiguation4/5

    Tools clearly distinguish between active and historical states, and different resource types (participants vs conferences vs nodes). However, the high number of similar get/list pairs (e.g., get_node_statistics vs get_node_status) could confuse agents, though descriptions help.

    Naming Consistency5/5

    Almost all tools follow a consistent verb_noun pattern with get_ and list_ prefixes in snake_case. Exceptions like summarize_calls and get_resource_schema are minor and still descriptive. No mixing of conventions.

    Tool Count2/5

    With 76 tools, the count far exceeds the 25+ threshold considered 'too many.' While the domain is complex, this many tools risks overwhelming agents and increases cognitive load. A more curated set or consolidated tools could improve usability.

    Completeness4/5

    The tool set covers a wide range of read and monitoring operations across conferences, participants, nodes, configurations, and history. The catch-all list_resources fills many gaps. Missing write operations (create/update/delete) are acceptable given the read-only intent, but a few minor read gaps exist (e.g., no get_alarm for active alarms).

    Maintenance

    ActivityInactive
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      D
      maintenance
      An MCP server that enables AI assistants to manage network infrastructure through the Equinix Fabric API, including ports, connections, and service profiles. It provides 22 tools for tasks like creating virtual connections, managing service tokens, and querying metro locations.
      1
      MIT
    • A
      license
      B
      quality
      D
      maintenance
      An MCP server that enables Claude to manage infrastructure across Kubernetes, Docker, Prometheus, and Terraform through natural language. It provides over 42 specialized tools with a safety-first design, including risk-based command classification and audit logging.
      43
      MIT