HaloPSA MCP Server

A Model Context Protocol server that lets AI agents on the MSPbots Agent Platform work in HaloPSA: tickets and ticket notes, clients, sites, contacts, assets, agents and teams, invoices, and the reference tables (statuses, priorities, ticket types, categories) needed to write tickets correctly.

It runs as a stateless HTTP service behind the MSPbots MCP gateway. The gateway sends each tenant's HaloPSA credentials as request headers on every call; the server keeps nothing between requests.

MSPbots Connector

The MCP Gateway built for MSPs: visible, controlled, and your data stays where it belongs.

Get started at connect.mspbots.ai →

MSPbots Connector gives your AI clients one MCP URL to reach your entire MSP stack (RMM, PSA, documentation, security, finance), and gives you control over who can call which tool and a record of everything that was called.

Hosted service: connect.mspbots.ai · Free to use.

Why Connector

Quick start

1. Sign up at connect.mspbots.ai.

2. Connect your vendors. Pick from 150+ connectors and authorize each one on the vendor's own page, or bring your own API keys. No extra vendor accounts needed.

3. Set access. Create roles and decide which tools each role may call. Nothing is allowed until you allow it.

4. Add your MCP URL to your AI client. Copy the URL from your Connector dashboard into Claude, ChatGPT, Cursor, VS Code, Gemini, Codex, or any MCP-capable client.

5. Ask your AI to do real work, and watch every call appear in the audit log.

Features

Unified gateway

• One MCP URL for all your AI clients

• 150+ connectors across the MSP stack

• Custom MCP servers supported, free

• Authorization happens on each vendor's own page; the AI never sees passwords or tokens

• Bring your own credentials: use the vendor API keys you already have

• Multiple accounts per app; connect once, reuse across AI clients

• Tool search: finds the right action for the task

• One sign-in; vendor credentials are injected behind the scenes and OAuth tokens refresh automatically

Visibility

• Audit log for every tool call (who, which vendor, when)

• Usage dashboard: call volume, top vendors, error rate, per-user activity

• Export audit logs to CSV, or stream to Loki, Graylog, or LogScale (SIEM)

• Optional recording of tool arguments and result summaries (off by default, enabled per organization)

Control

• Per-role, per-tool allowlist, deny by default

• owner / admin / member roles

• SCIM provisioning with Entra ID, Okta, JumpCloud, and Google Workspace

Security

• Credentials encrypted one record at a time (AES-256-GCM, per-record key)

• AI clients authenticate with OAuth 2.1 + PKCE

• Vendor responses are scanned for prompt-injection patterns and returned as structured data. This is one layer of defense, not a guarantee against every attack.

• Static egress IP for customer firewall and API allowlisting

Pricing

Free.

Links: connect.mspbots.ai · mspbots.ai

Related MCP server: HaloPSA API Gateway MCP Server

Credentials (request headers)

Header names are case-insensitive. A request to /mcp without the client ID, the secret, or both the tenant and base URL gets 401 with the list of expected headers. The middleware only checks that the headers are present; a malformed tenant or base URL is reported by each tool as not_configured.

The server gets an OAuth2 token (client_credentials, scope=all) from <instance>/auth/token on every tool call and calls the REST API at <instance>/api. When both a base URL and a tenant are sent, the tenant is added to the token request as ?tenant= (as HaloPSA requires for some hosted instances). Tokens are never cached between requests. Give the API application an agent with the permissions the agents need (read tickets, clients, assets, invoices; write tickets, actions, clients and contacts).

Tools

Tool names and parameters match the previous HaloPSA connector on the MSPbots Agent Platform, so existing agents keep working. This is why the server has more than 20 tools.

Behaviour notes:

• Every tool returns a JSON string. Errors use one envelope: {"error": {"code": "...", "message": "...", "retryable": false}} with code one of not_configured, unauthorized, not_found, invalid_argument, rate_limited, upstream_error. HaloPSA's own error text is included in the message.

• List tools return {"items", "count", "limit", "has_more"} (plus total_count / page_no where HaloPSA reports them). limit defaults to 50 (25 for the search tools, 100 for asset types) and is capped at 200; tickets are capped at 100 per page (HaloPSA's maximum) and paged with page_no. Results larger than 20,000 characters are cut and flagged with truncated / has_more.

• All tools are always listed. halopsa_navigate only describes the tools of one area (teams are under agents); halopsa_status shows which headers were sent, the instance URL and whether a token could be obtained, without revealing secrets.

• Tickets: HaloPSA expands related records (ticket type, priority, user) in full on a ticket. halopsa_tickets_get/create/update reduce those to {id, name} unless full_response is true.

• halopsa_tickets_update changes only the fields given. team_id assigns the team (the team name is looked up and sent with it); custom_fields writes custom fields by id or name; tickettype_id is applied through a hidden action and the ticket is re-read to confirm the change stuck, otherwise the tool returns an error.

• halopsa_tickets_add_action takes timetaken in minutes and sends HaloPSA hours.

• halopsa_users_create refuses (and returns the existing contact) when the email is already used by any contact, active or inactive, and defaults site_id to the client's main site.

• halopsa_invoices_list: status accepts posted / notposted (sent to HaloPSA); sent, paid and the invoice date range are applied to the latest 200 invoices HaloPSA returns and the result says so.

• halopsa_lookups_get returns the identifying fields of each row; for priorities, ticket_priority_id is the value a ticket's priority_id takes. A table that fails is reported in place without failing the others.

• Transient failures (429, 5xx, timeouts) are retried with backoff, honouring Retry-After; one tool call stays under 120 seconds.

Run locally

uv sync
uv run halopsa-mcp            # listens on 0.0.0.0:8080
curl -s http://127.0.0.1:8080/health
curl -s http://127.0.0.1:8080/mcp \
  -H 'Content-Type: application/json' -H 'Accept: application/json, text/event-stream' \
  -H "X-Halo-Client-Id: $HALO_CLIENT_ID" -H "X-Halo-Client-Secret: $HALO_CLIENT_SECRET" \
  -H 'X-Halo-Tenant: yourcompany' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

Environment variables (all optional): MCP_HTTP_HOST (default 0.0.0.0), MCP_HTTP_PORT (default 8080), MCP_TRANSPORT (http or stdio). Unknown variables are ignored. Credentials are never read from the environment.

Docker

docker build -t halopsa-mcp .
docker run --rm -p 8080:8080 halopsa-mcp

Development

uv sync
uv run pytest -q
uv run ruff check . && uv run ruff format --check .

Code is in src/halopsa_mcp/ (one module per area under tools/), tests in tests/. The implementation follows the public HaloPSA API documentation at https://halo.halopsa.com/apidoc/ (each instance also serves it at /apidoc/).

License

Copyright 2026 MSPbots.ai. Licensed under the Apache License, Version 2.0. See LICENSE and NOTICE.

HaloPSA is a trademark of its owner. This project is not affiliated with or endorsed by Halo.