Loading ToolBench
Local MCP server exposing the Ringg AI platform API over stdio.
20 tools with complete JSON Schema definitions and descriptions. Naming follows verb_noun convention consistently (list_, get_, update_, attach_, detach_, toggle_). Descriptions are present but brief (avg ~60 chars, below the 194-char baseline). All parameters have type definitions and descriptions. However, output schemas are not documented, LLMs cannot predict response structure. Error handling is basic (no recovery guidance, no categorization). No tool annotations (readOnlyHint/destructiveHint) despite clear risk stratification (10 READ_ONLY, 10 WRITE). Missing idempotency guidance for write tools.
Add a new A/B testing version to an agent
Attach a knowledge base to an agent
Detach a knowledge base from an agent
Get detailed information about a specific agent
Get detailed information about a specific call
Get detailed information about a specific knowledge base
List agents with pagination
Output schemas not documented. LLMs cannot predict response structure or plan downstream tool calls. E.g., list_agents returns what fields? Does it include pagination metadata?
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). 10 WRITE tools lack destructiveHint; LLMs cannot distinguish safe from risky operations without explicit marking.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | <=2025-11-25 | v2 |
List call history with optional filtering
List all knowledge bases
Enable or disable A/B testing for an agent
Transcribe audio using Ringg's Parrot speech-to-text service
Update the agent's display name
Update one or more prompt sections of an agent
Update analytics context for an agent
Update classification labels for an agent
Update client analysis configuration for an agent
Update the custom analysis prompt for an agent
Update custom variables for an agent
Update the agent's intro message
Update traffic split between A/B versions
Descriptions are brief (avg ~60 chars, well below 194-char baseline). Missing context on WHEN to use each tool and what it returns. E.g., 'Update one or more prompt sections of an agent' does not explain what 'sections' are or what happens to existing sections.
No error recovery guidance. Error responses do not tell LLMs what to do next (retry, ask user, try alternative tool). E.g., if transcribe_audio fails, should the agent retry or suggest a different format?
No idempotency guidance for write tools. Agents retry on ambiguous failures, update_agent_prompt called twice with same args should be safe, but this is not documented.