Loading ToolBench
MCP server for multi-agent task management via tasks.md
vibe-conductor has 7 well-named tools with clear verb-noun patterns (list_tasks, get_task, claim_task, etc.). All tools have descriptions (avg ~80 chars, within baseline 34-392). Input schemas are present with type definitions. However, parameter descriptions are sparse or missing context. Output schemas are not documented. Error handling lacks recovery guidance. No tool annotations (readOnlyHint/destructiveHint). Composition is sound, tools chain well (list → get → claim → complete). Security: no secrets in params, but no explicit permission gates or audit trail declarations.
Append notes to a task body (e.g. changed files, findings, context)
Mark a task as blocked when you cannot proceed. The human will be notified.
Atomically claim a todo or commit task. Prevents double-claiming. Only one commit task may be active at a time.
Mark a task as complete. Behavior depends on task type (implement→review, test→review, commit→done)
Create a new task in a specified section with optional body and metadata
Get full details of a task including body and notes
List tasks with optional filtering by status, session, section, or priority
Output schemas not documented. LLMs cannot infer what fields list_tasks, get_task, and create_task return, forcing them to guess downstream field names and risk broken tool chains.
Parameter descriptions lack actionable constraints. 'status' param in list_tasks lists enum values in description but not as formal enum constraint. 'priority' is boolean but description doesn't clarify what true/false means. 'section' accepts paths like 'Core > Parser' but format is not documented.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). LLMs cannot distinguish safe reads from destructive writes without explicit hints. claim_task and complete_task are marked WRITE but lack destructiveHint annotation.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 60 | <=2025-11-25 | v2 |
Error handling lacks recovery guidance. No evidence of actionable error messages (e.g., 'Task not found. Available tasks: [list]' or 'Cannot claim: task already claimed by session X'). Errors likely return generic failures.
add_task_note has ambiguous parameter design: both 'note' (string) and 'lines' (array) are present. Unclear which takes precedence or if both can be used together. Undocumented parameter relationships.