Loading ToolBench
Scaffold and maintenance CLI for MaaFW application projects
Server provides 9 tools with basic schemas and descriptions, but quality is inconsistent. Tool names follow verb_noun convention (create_project, sync_project, run_doctor, clean_cache, list_project_backups, inspect_project_backup, restore_backup, record_update_requests, apply_incremental_addons). Descriptions exist but are terse (10-60 chars), below the 50-200 char LLM-optimized baseline. Input schemas are present with type declarations, but parameter descriptions are minimal or missing context about constraints, formats, and dependencies. No output schemas documented. Error handling guidance absent. Security concerns: tools accept 'root' directory paths without visible sanitization against path traversal. No permission gates on destructive operations (clean_cache, restore_backup). Composition is reasonable, tools are single-purpose, but chaining support is unclear (no documented ID/reference flow between tools).
Apply add-ons incrementally to an existing project
Clean the project cache directory
Create a new MaaFW application project with scaffolding
Inspect details of a project backup
List available project backups
Record requests to update project components
Restore a project from a backup
Run diagnostic checks on a MaaFW project
Synchronize an existing MaaFW project with latest templates and dependencies
Descriptions are too terse (10-60 chars vs. 50-200 char baseline). LLMs cannot determine when to select tools or understand prerequisites. E.g., 'Create a new MaaFW application project with scaffolding' lacks context on when to use vs. sync_project, what scaffolding includes, or what happens to installNodeDeps/downloadOcrModels.
No output schemas documented. LLMs cannot plan downstream calls or extract required fields. E.g., does create_project return a project_id, path, or status? Does list_project_backups return backup metadata needed by inspect_project_backup?
Path traversal risk: 'root' parameter accepts arbitrary directory paths with no visible validation. Agents could be tricked into operating on sensitive directories outside the project. Sanitize against '..' and absolute paths; validate root is a valid MaaFW project.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 44 | <=2025-11-25 | v2 |
Destructive operations (clean_cache, restore_backup) lack permission gates, confirmation steps, or dry-run support. An agent could irreversibly delete cache or restore wrong backups without safeguards. Implement confirmation_request pattern.
Parameter descriptions are minimal. E.g., 'update' in sync_project is 'List of components to update', what components? What format? What are valid values? 'addons' in apply_incremental_addons lacks any description. LLMs cannot infer valid inputs.