Loading ToolBench
MCP server for Convertigo Flow, enabling Flow-native project authoring and integration with Codex AI assistants
This MCP server has critical definition gaps. Tool names lack action verbs (flow-apply, codex-setup, frontend-asset-import are vague). Descriptions are present but generic (10-50 chars for most). Input schemas exist but are incomplete: many parameters lack type constraints, enums, or validation guidance. Output schemas are largely undocumented. Error handling is absent, no recovery guidance, no categorization. The server appears to be a Convertigo Flow integration layer, but the tool interface does not follow agentic tool patterns. Tool composition is unclear: frontend-svelte-source has 15+ parameters with overlapping concerns (get, rg, check, set, patch operations should be separate tools). No evidence of idempotency, pagination, or chaining IDs in responses.
Installs or updates local Codex onboarding for Convertigo Flow MCP.
Runs the flow-apply MCP tool.
Imports one local image into the target project's resources directory and returns its canonical resources/... URL.
Learns a safe requestable schema and attaches it to one FullSync frontend action.
Reads, searches, validates and writes Flow Svelte models, provider components or canonical Flow CSS sources.
Tool names lack action verbs. 'flow-apply', 'codex-setup', 'frontend-asset-import' do not start with clear verbs (apply_flow_mutation, setup_codex_onboarding, import_frontend_asset). LLMs cannot infer intent from these names.
frontend-svelte-source combines 5 operations (get, rg, check, set, patch) in one tool with 15+ parameters. This violates single-responsibility principle. Should be split into separate_tools: get_svelte_source, search_svelte_source, validate_svelte_source, update_svelte_source, patch_svelte_source.
Output schemas are not documented. No tool shows what fields are returned or their types. LLMs cannot plan downstream calls or extract data without knowing response structure.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 27 | 2025-06-18+ | v2 |
Parameters lack type constraints and validation guidance. 'operation' in frontend-svelte-source should be an enum (get|rg|check|set|patch). 'pattern' lacks regex/format guidance. 'limit' (1-200) is documented but not enforced in schema.
No error handling or recovery guidance. Tools do not document what errors can occur, whether they are retryable, or what the LLM should do next. No dry-run or confirmation patterns for destructive operations (write, delete, overwrite).