Related Servers

Every tool has a clearly distinct purpose, and the descriptions actively cross-reference each other to prevent misselection: deploy_mock_from_* explicitly warns it is NOT valid input for serve_locally, request_history vs diagnose_requests are differentiated as list-vs-diagnose, and wait_for_deploy vs wait_for_github_deploy are separated by deployment target. Even the three near-identical deploy_mock_from_catalog/spec/url tools are cleanly distinguished by input source. With 35 tools this is an exceptional level of disambiguation.

The overwhelming majority follow a consistent snake_case verb_noun pattern (list_replays, serve_locally, deploy_mock_from_catalog, wait_for_github_deploy, unpublish_from_github), and related families share structural prefixes (mockzilla_docs_*, deploy_mock_from_*, wait_for_*). Minor deviations include bare verbs (info, lint, simplify, pack) and a noun-first name (bridge_status), but these are isolated and the overall pattern remains highly predictable.

35 tools is on the heavy side and exceeds the calibration's 25-tool threshold for 'too many', but the domain is genuinely broad: local serving, hosted deployment, GitHub publishing, replay, spec processing, CLI management, and docs each demand their own surface. Some consolidation is possible (deploy_mock_from_spec and deploy_mock_from_url could be one tool; wait_for_deploy vs wait_for_github_deploy) but each tool does earn a place in the platform's scope.

The surface covers the full lifecycle remarkably well: create (serve_locally, mock_endpoint, deploy_*), read (list_sims, list_mock_endpoints, request_history, info), modify (setup_replay, simplify), and delete (stop_locally, clear_mock_endpoints, unpublish_from_github). Workflows are chained explicitly (deploy → wait_for_deploy → list_sims). The only notable gap is the absence of update/delete operations for hosted sims themselves — list_sims lists them but nothing can remove or modify an individual hosted sim.