Loading ToolBench
A local, private task manager and document system with MCP support for assistants
Tisty exposes 30 tools via STDIO with basic schemas and descriptions. Most tools have descriptions (10-100 chars, below the 50-200 char baseline for LLM optimization), and input schemas are present but sparse. Many parameters lack type constraints (enums, ranges). Critical gaps: no output schemas documented, no error recovery guidance, no tool annotations (readOnlyHint/destructiveHint), and parameter descriptions are often generic ('Task selector', 'Filter criteria'). The 'set' tool conflates 10+ responsibilities (title, date, deadline, priority, tags, repeat, reminder, read_as) into one, violating single-responsibility. Tools like 'sync', 'doctor', 'attach' have vague descriptions that don't explain when to use them or what they return. No evidence of idempotency guarantees or confirmation patterns for destructive ops (rm, leave). Naming is mostly verb-noun (good), but 'mcp' and 'agent' are cryptic.
Add a new task with optional date, deadline, priority, and list assignment
Let an assistant file work here, or stop it
Keep a copy of a file with a task or in a document
Manage configuration (get, set, unset, path)
Create demo data for testing
Set or clear task description
List documents, read one, or write one with --new
Tool 'set' violates single-responsibility principle: conflates title, date, deadline, priority, tags, repeat, reminder, and read_as into one tool. Should split into set_title, set_date, set_priority, add_tags, set_repeat, set_reminder.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., task_id after 'add'). Baseline: 100% of A+ tools document return types.
Parameter descriptions are generic and lack constraints. 'Filter criteria' (ls), 'Task selector' (done, undone, drop, rm, show, story, step), 'Log text' (log) do not explain format, valid values, or examples. Baseline: 100% of A+ tool params have descriptions; 94% of A+ tools have docstrings.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 46 | <=2025-11-25 | v2 |
Check and optionally repair data integrity
Mark a task as done, optionally with additional dates
Drop a task (soft delete)
Export tasks with optional markdown format
Take Tisty's own settings, cache and command line off this machine. Your documents, tasks and attachments stay where they are.
Manage lists (ls, add, rename, archive, unarchive, rm)
List all task lists
Add a log entry to a task
List tasks with optional filters
Speak MCP over stdin and stdout, so an assistant can file work here
Move a task to a different list or inbox
Redo the last undone change
Remove a task permanently
Search tasks by query, optionally filtering by open/archive status
Show task series/recurring instances
Set task properties like title, date, deadline, priority, tags, repeat, and reminder
Show detailed information about a task
Manage task steps (add, mark done/undone, edit text, remove)
Show task story/narrative view
Synchronize data with backup or other machines
Manage tags (ls, rename, rm)
Undo the last change
Mark a task as not done
No error recovery guidance. Tools like 'sync', 'doctor', 'attach' lack descriptions explaining when to use them, what they return, or how to handle failures. No actionable error messages documented.
Destructive tools (rm, leave) lack confirmation or dry-run patterns. No evidence of idempotency guarantees or permission gates. Agents could accidentally delete data without safeguards.