Loading ToolBench
A visual workflow engine with MCP server capabilities. Allows users to create, validate, and execute workflows as tools exposed via MCP servers. Includes a web-based editor, credential management, data tables, vector stores, and multi-user support.
LibreFlow exposes 19 tools with consistent naming (verb_noun pattern) and basic descriptions. All tools have input schemas with type definitions. However, descriptions are terse (avg ~60 chars, well below the 194-char baseline), parameter descriptions are minimal or absent, and output schemas are not documented. Error handling guidance is missing. No tool annotations (readOnlyHint/destructiveHint) despite clear risk classifications in the metadata. Composition is sound, tools are single-responsibility and chainable via IDs, but LLM-facing descriptions need expansion to meet production standards.
Add a new row to a data table
Create a new data table with specified columns
Delete an entire data table
Delete a row from a data table
Delete a workflow
Query rows from a data table with pagination
Get a specific workflow execution by ID
Descriptions are terse (avg ~60 chars vs 194-char baseline). Most lack context on WHEN to use the tool, prerequisites, or what the LLM should expect. E.g., 'List all workflows accessible to the current user' does not explain filtering, pagination, or return structure.
Output schemas are not documented. LLMs cannot infer what fields are returned, their types, or which IDs/references are available for chaining. E.g., libreflow_list_workflows returns a list but structure is opaque.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 45 | 2026-07-28+ | v2 |
Get a specific workflow by ID
Atomically increment a numeric field in a data table row
List all data tables accessible to the current user
List workflow executions
List all available node types that can be used in workflows
List all workflows accessible to the current user
Execute a workflow and return the execution report
Create or update a workflow
Enable or disable a workflow
Update an existing row in a data table
Insert or update a row in a data table (idempotent on key column)
Validate a workflow for coherence and correctness
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk classifications in metadata (READ_ONLY, WRITE, DESTRUCTIVE). LLMs cannot distinguish safe reads from destructive writes without explicit hints.
Parameter descriptions are minimal or absent. E.g., 'nodes' in libreflow_save_workflow is described as 'Array of workflow nodes' but lacks structure, required fields, or validation rules. LLMs cannot construct valid payloads.
No error handling guidance. Tools do not document what errors are retryable, what user-fixable errors look like, or how to recover. E.g., libreflow_run_workflow may fail if the workflow is invalid, but no guidance on next steps.