Orchestrator of skills and tools for AI agents. Manages tool execution, MCP process pooling, categories, runners, and execution logging.
ToolHub exposes 20 tools with basic schemas and descriptions, but exhibits critical gaps in naming clarity, parameter documentation, and error handling. Tool names lack consistent verb-noun structure (e.g., 'listTools' vs 'callTool' vs 'getSmartPrompt'). Most parameter descriptions are minimal (10-30 chars), well below the 72-char baseline. No output schemas documented. Destructive operations (deleteCategory, deleteTool, deleteRunner, clearExecutionLogs) lack confirmation patterns or dry-run support. Error handling is absent from visible code. Security concerns: updateSystemSettings accepts agentSecret and adminPassword as parameters, violating secret-injection pattern. No audit logging visible. Tool composition is poor: callTool accepts generic 'payload' object with no schema, forcing LLMs to guess structure.
Execute a tool at a given path with provided payload
Clear all execution logs
Create a new tool category
Create a new execution runner (environment)
Create a new tool with code, runner, and metadata
Delete a tool category
Delete a runner
Delete a tool
callTool accepts untyped 'payload' object with no schema. LLMs cannot infer required fields, types, or structure. Forces guessing and trial-and-error invocations.
Destructive operations (deleteCategory, deleteTool, deleteRunner, clearExecutionLogs) lack confirmation patterns, dry-run support, or recovery guidance. Agents can irreversibly delete data without safeguards.
updateSystemSettings exposes agentSecret and adminPassword as tool parameters. Secrets in parameters leak into logs and prompt history. Violates secret-injection pattern.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields (e.g., category_id after createCategory). Breaks tool chaining.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | <=2025-11-25 | v2 |
Export tools and categories as a .toolpack file
Retrieve execution logs with optional filtering
Get status of MCP process pool
Retrieve the system prompt for AI agents, combining root prompt and append prompt
Retrieve system settings
Import tools and categories from a .toolpack file
Kill a specific MCP process
List available tools and categories at a given path in the tool hierarchy
Update an existing tool category
Update an existing runner
Update system settings
Update an existing tool
Parameter descriptions are minimal (10-40 chars, well below 72-char baseline). Missing format constraints, ranges, and dependencies. E.g., 'config' in createRunner has no guidance on valid keys or structure.