Connect Claude Desktop, Cursor, and other AI tools to FlowDot workflows - create, execute, and manage visual automation workflows
Scoring was not performed
Output schemas completely undocumented for all 14 tools. LLMs cannot predict response structure, breaking tool chaining and forcing exploratory calls.
Error handling not documented. Tools do not explain recovery paths (e.g., 'if app_id not found, try list_apps first') or error classification (retryable vs user-fixable).
Parameter constraints lack specificity. E.g., edit_app_code accepts 'replace_all' boolean but does not document what happens if old_string appears multiple times and replace_all=false. Code size limits (102KB mention in description) not tied to validation.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 22 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 49 | - | v1 |
Example code in descriptions may cause literal reuse. Descriptions for create_app and edit_app_code include React code examples that LLMs may copy-paste rather than adapt.
Tool 'update_app_code' listed but not visible in provided source code. Cannot verify definition.