No user-submitted related servers found.
Scored across 40 tools
Descriptions are unusually explicit about boundaries and cross-reference the correct sibling tool, which rescues much of the set. However, there are several near-parallel families — terminal_read vs terminal_wait, terminal_read_activity vs terminal_wait_activity, terminal_read_device_events vs terminal_wait_device_event, terminal_session_events, terminal_read_diagnostic — plus terminal_write vs terminal_write_leased vs terminal_exec, and terminal_acquire_lease vs terminal_renew_lease vs terminal_release_lease, giving real misselection risk across 40 tools.
Consistent snake_case with predictable terminal_ / debug_ / debug_bridge_ prefixes that map to sub-domains, which is a sensible convention at this scale. Minor variance in element ordering (some verb-first like terminal_get_connection_decision and terminal_list_serial_ports, others noun-first like terminal_session_events, terminal_file_transfer_checkpoint) keeps it from a perfect score.
40 tools is well past the point of easy selection for an agent, and a substantial share are structural duplicates (paired immediate-read / wait variants) or lease bookkeeping rather than distinct capabilities. The domain is genuinely complex (serial sessions, U-Boot entry, GDB bridge, file transfer, diagnostics), but that breadth is used to justify a surface that is too large.
Lifecycle coverage is strong: session open/close/attach/detach, lease acquire/renew/release, read/write, device discovery and events, U-Boot entry, native debug start/stop/command/events, debug bridge start/status/renew/report/restore/close, plus file-transfer checkpoint/cancel and diagnostics export/download. Gaps are minor and documented — no tool to create or edit serial profiles (terminal_open_serial assumes one), and file upload/download is pushed to the CLI rather than the tool surface.