Trigger.dev MCP by usefulapi

Each tool targets a distinct resource+action, and the run-oriented cluster (get_run, get_run_metadata, get_run_trace, get_run_span, list_run_events, list_runs) is cleanly separated by clear suffixes and complementary descriptions. Schedule vs queue vs run vs batch operations are unambiguous, and pause_queue vs deactivate_schedule are distinct resources.

Every tool follows the same trigger_verb_noun snake_case pattern (trigger_get_run, trigger_list_schedules, trigger_cancel_run, trigger_pause_queue). No mixing of conventions or vague verbs; the pattern is fully predictable.

22 tools is on the heavy side but appropriate for a full platform API spanning runs, schedules, queues, batches, deployments and env vars. Each tool maps to a distinct endpoint and earns its place, though the surface is dense enough to feel slightly sprawling.

Observability is thorough (runs, traces, spans, events, metadata, queues, deployments), but key lifecycle operations are missing: no tool to trigger/submit a new run, no create/delete for schedules, and bulk actions are list-only with no way to initiate one. Env vars are explicitly read-only. These gaps limit the server to monitoring and control of existing runs rather than full lifecycle coverage.