Loading ToolBench

Reading the current corpus…

mcp-terminator MCP server · D · ToolBench

← Browse

mcp-terminator

github.com/davidroman0O/mcp-terminator

MCP server for terminal session management with UI element detection and terminal automation

2026-07-28+STDIO10 tools

56/100

Definition Quality76

Protocol Readiness50

Supportability26

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence

Current-spec patterns detected

Summary

Server has 10 well-named tools with complete input schemas and descriptions. Naming follows verb_noun convention (terminal_session_create, terminal_read_output). All tools have descriptions (avg ~60 chars, within baseline). Input schemas are properly typed with JSON Schema. However, output schemas are not formally documented, responses are inferred from jsonResult() calls. Error handling is minimal (generic toolError messages without recovery guidance). No parameter constraints (enums, ranges) despite many accepting free-form strings. Missing idempotency hints and permission declarations. Composition is sound (each tool has one responsibility), but descriptions lack WHEN/WHY context that LLMs need for tool selection.

Tools (10)

terminal_clickwritesource verified72/100

Click at a position in the terminal session

terminal_press_keywritesource verified75/100

Press a key or key combination in the terminal session

terminal_read_outputread onlysource verified77/100

Read raw terminal output since the last read

terminal_session_closedestructivesource verified77/100

Close and terminate a terminal session

terminal_session_createwritesource verified82/100

Create a new terminal session with the specified command

terminal_session_listread onlysource verified75/100

List all active terminal sessions

terminal_session_resizewritesource verified77/100

Resize a terminal session to new dimensions

terminal_snapshot

Issues (5)

HIGHtool ↗

Output schemas not formally documented. Responses are constructed via jsonResult() helper but no schema definitions exist in tool registration. LLMs cannot predict return structure for planning downstream calls.

terminal_session_createterminal_session_listterminal_session_closeterminal_session_resizeterminal_snapshotterminal_read_outputterminal_typeterminal_press_keyterminal_clickterminal_wait_for

HIGHrecovery-guide ↗

Error messages are generic ('failed to create session: %v'). No recovery guidance or error classification (retryable vs fatal). LLMs cannot determine next action on failure.

terminal_session_createterminal_session_close

MEDIUMconstrained-input ↗

Parameter descriptions lack constraints. 'key' in terminal_press_key accepts free-form strings (e.g., 'Enter', 'Ctrl+c', 'Up') but no enum or pattern is declared. LLMs may hallucinate invalid key names.

terminal_press_key

Recommendations

Add formal output schemas to each tool via mcp.WithObject() or equivalent. Document return fields (session_id, status, output, etc.) with types and descriptions so LLMs can plan chaining.
Enhance error messages with recovery guidance. E.g., 'Session not found (ID: xyz). Call terminal_session_list() to see active sessions.' Classify errors as retryable or fatal.
Declare enums for constrained parameters: terminal_press_key's 'key' should list valid values (Enter, Escape, Tab, Ctrl+c, Ctrl+d, Up, Down, Left, Right, etc.). Use mcp.WithEnum() or add pattern/enum to schema.
Add tool annotations: mark terminal_session_close, terminal_type, terminal_press_key, terminal_click as destructiveHint=true; mark terminal_session_list, terminal_read_output, terminal_snapshot as readOnlyHint=true.
Expand descriptions to 100-150 chars with WHEN/WHY context. E.g., 'Type text into the terminal session. Use this to send input to interactive programs (editors, REPLs). Returns the number of characters typed. For key combinations, use terminal_press_key instead.'

Score history

Overall score trend

First recorded score · v2 rubric

56/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	D	56	2026-07-28+	v2