An MCP server for controlling Zellij terminal multiplexer via Model Context Protocol. Provides tools for managing sessions, tabs, panes, and terminal I/O operations.
19 tools with consistent naming (verb_noun pattern) and basic descriptions. However, descriptions are uniformly terse (10-30 chars), falling below the 50-200 char LLM-optimized baseline. Input schemas are present and typed, but lack parameter descriptions entirely, a critical gap. No output schemas documented. Error handling and recovery guidance absent. Tool composition is sound (single responsibility), but parameter constraints (enums, ranges) are missing. Security considerations (secret injection, permission gates) not evident.
Attach to a specific pane
Attach to an existing Zellij session
Close the current pane
Close the current tab
Create a new Zellij session
Detach from the current Zellij session
Dump the current layout of the session
Dump the current screen content from the pane
Parameter descriptions completely absent. All 19 tools have typed input schemas but zero parameter-level documentation. LLMs cannot infer meaning from names alone (e.g., 'range' in dump_screen could mean line numbers, byte offsets, or viewport coordinates).
Tool descriptions are uniformly terse (10-30 chars), well below the 50-200 char LLM-optimized baseline. Descriptions lack context on WHEN to use each tool, prerequisites, and what the tool returns. Example: 'List all active Zellij sessions' does not explain when to call this vs attach_session or what the response structure is.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 46 | <=2025-11-25 | v2 |
List all panes in the current tab
List all active Zellij sessions
List all tabs in the current session
Create a new tab in the current session
Rename the current pane
Rename the current Zellij session
Rename the current tab
Send a key press to the current pane
Switch to a specific tab in the current session
Write multiple lines of text to the current pane
Write text to the current pane
No output schemas documented. LLMs cannot plan downstream tool calls or extract required fields (e.g., session IDs from list_sessions, pane IDs from list_panes) without knowing response structure. This breaks tool chaining.
No error handling or recovery guidance. Tools like close_tab and close_pane (DESTRUCTIVE risk) have no confirmation step, dry-run option, or error messages guiding the LLM on what to do if the operation fails.
Missing parameter constraints. 'key' in send_key accepts free-form strings with no enum or format guidance. 'range' in dump_screen has no documented format (is it 'last:10', 'range:0:50', 'viewport'?). LLMs will hallucinate invalid values.