{"id":"wyre-technology-autotask-mcp","name":"autotask-mcp","af_score":59.2,"security_score":53.8,"reliability_score":35.0,"what_it_does":"Provides an MCP (Model Context Protocol) server that connects an AI assistant to Kaseya Autotask PSA, exposing ~39 tools for common MSP workflows such as searching/creating/updating companies, contacts, tickets, time entries, projects, billing items, notes, and attachments, with optional HTTP Streamable transport and an MCP bundle for desktop clients.","best_when":"You run an MSP that already has Autotask API access and want an MCP-enabled AI assistant to execute or assist with day-to-day PSA operations using either local stdio or a Docker-hosted HTTP transport.","avoid_when":"You cannot restrict outbound/inbound network access, or you cannot safely manage Autotask credentials used by the MCP server.","last_evaluated":"2026-03-30T15:28:46.754552+00:00","has_mcp":true,"has_api":false,"auth_methods":["Autotask credentials via environment variables (AUTOTASK_USERNAME, AUTOTASK_SECRET, AUTOTASK_INTEGRATION_CODE)","Gateway mode header injection (X-API-Key, X-API-Secret, X-Integration-Code, optional X-API-URL)"],"has_free_tier":false,"known_gotchas":["Create/update tools may have side effects; without documented idempotency and retry guidance, agents should avoid blind retries after partial failures.","Gateway mode relies on per-request credential headers; incorrect header handling or missing transport security could expose credentials.","Autotask API behavior (rate limiting, partial failures, and eventual consistency) is not documented here; agents should expect Autotask-specific constraints."],"error_quality":0.0}