Loading ToolBench
Gong MCP server — a full-surface Model Context Protocol gateway to the Gong REST API (calls, transcripts, users, activity, scorecards, coaching, CRM).
23 tools with consistent naming (verb_noun pattern) and basic schemas. All tools are READ_ONLY and properly typed. However, descriptions are minimal (10-80 chars, well below the 194-char baseline), parameter descriptions lack depth, and output schemas are undocumented. No error handling guidance, no pagination documentation despite max_pages parameters, and no chaining IDs in responses. Tool definitions are visible in source but lack LLM-optimization for agent selection.
Get details of a specific call by ID
Get details of a specific coaching account by ID
Get details of a specific coaching opportunity by ID
Get details of a specific CRM account by ID
Get details of a specific CRM opportunity by ID
Get content of a specific library folder
Get details of a specific scorecard by ID
Tool descriptions are 10 - 80 characters, far below the 194-char baseline. Descriptions like 'List all workspaces' and 'Get details of a specific call by ID' lack context for LLM selection. No guidance on WHEN to use each tool vs. similar alternatives (e.g., list_calls vs. list_calls_extensive).
Output schemas are completely undocumented. No indication of what fields are returned, their types, or which IDs/references are included for chaining. LLMs cannot plan downstream calls or extract relevant data.
Pagination parameters (max_pages, cursor handling) are present but not documented in tool descriptions. No guidance on pagination limits, result counts, or when pagination is required. Agents cannot reason about result sizes.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 48 | 2026-07-28+ | v2 |
Get transcripts for calls with optional filtering
Get details of a specific user by ID
Get extensive user data filtered by user IDs or email addresses
List activity records with optional filtering
List scorecards for a specific call
List calls with optional filtering by date range and workspace
List calls with extensive content and interaction data including transcripts, highlights, topics, trackers, and more
List coaching accounts
List coaching opportunities with optional filtering
List CRM accounts with optional filtering
List CRM opportunities with optional filtering
List library folders in a workspace
List questions for a specific scorecard
List scorecards with optional filtering
List users with optional avatar inclusion
List all workspaces
No error handling guidance. No indication of what errors are retryable, what user-fixable errors look like, or how to recover from API failures. Agents receive raw error messages with no actionable next steps.
Parameter descriptions are minimal (5 - 30 chars). E.g., 'Start datetime for call filtering' lacks format guidance (ISO 8601?), timezone info, or range constraints. LLMs cannot reliably construct valid inputs.