Loading ToolBench
An MCP server that orchestrates a crew of agentic coding agents over the Agent Client Protocol (ACP).
Critical failures across naming, descriptions, and schemas. Of 18 tools, only 3 have visible input schemas (delegate, setup, job_status). Most tools (15/18) show empty input objects {} with no parameter documentation. Descriptions are present but generic (avg ~40 chars, well below 194-char baseline). Tool names lack clear verb-noun structure (e.g., 'consensus', 'debate', 'discover' are vague). No output schemas documented. Error handling and recovery guidance absent. This server prioritizes orchestration complexity over agent-usable tool design.
Get the focused snapshot of just the jobs in flight right now.
Analyze code or output.
Cancel a background job.
Show which agents are available and their capabilities.
Build consensus across multiple agents on a task.
Continue a paused background job.
Run a debate across multiple agents.
15 of 18 tools have empty input schemas ({}). No parameter types, descriptions, or constraints visible. LLMs cannot determine what inputs these tools accept or how to invoke them.
Tool names lack clear verb-noun structure. 'consensus', 'debate', 'discover', 'analyze', 'activity', 'reload_panels' are vague or generic. LLMs cannot infer intent from names alone. Should be: 'build_consensus', 'run_debate', 'find_agents', 'analyze_code', 'get_activity', 'refresh_panels'.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 41 | <=2025-11-25 | v2 |
Delegate a task to one ACP agent and return its normalized result.
Find ACP agents you already have installed and propose config for them.
Diagnose the Rutherford installation and agent setup.
Get the result of a completed background job.
Get the status of a background job.
List background jobs.
List available agent roles.
Create a plan for a task.
Reload named multi-agent panels from panels.toon.
Review work from an agent.
Show where config lives and scaffold a starter config.toml; with write create it (never clobber).
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields. Responses are unstructured, forcing LLMs to parse free-text output.
Descriptions are generic and under 50 chars (avg ~35 chars vs 194-char baseline). 'Diagnose the Rutherford installation' does not explain WHEN to call it, WHAT it returns, or HOW to act on results. No recovery guidance for errors.
No error handling or recovery guidance visible. Tools do not document what errors can occur, whether they are retryable, or what the LLM should do next. Agents will fail silently or retry indefinitely.