Loading ToolBench
MCP Server generated from OpenAPI spec for adblast-api-documentation. Provides tools to interact with the AdBlast API for managing projects, scoring, and video generation configurations.
This MCP server has 14 tools with schemas visible and largely complete parameter definitions. However, there are systematic quality issues that prevent a higher score: (1) tool descriptions are minimal and often lack context about when to use a tool vs similar alternatives; (2) parameter descriptions are present but frequently generic or lack constraints/validation guidance; (3) output schemas are not documented, the code shows input schemas clearly but never articulates what each tool returns; (4) error handling guidance is absent, no recovery hints or classification of errors; (5) no tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk profiles; (6) parameter naming inconsistencies (e.g., requestBody wrapper adds nesting complexity). The server is functional but falls short of production-grade tool design. Average across all 14 tools is 58; most score 55-65, with no tools reaching 75+.
Create Project
Create and generate the score configuration
Delete Project
Reset the score configuration to the default state
Retrieve the full project object
Retrieve the project's JSON data
Retrieve the JSON schema for project data validation
No output schemas documented. Tools declare input schemas but never specify what fields or structure they return. LLMs cannot plan downstream calls or extract relevant data without knowing response structure.
Tool descriptions lack actionable context. Examples: 'List Projects' (10 chars), 'Create Project' (14 chars), 'Delete Project' (14 chars), 'Update Project' (14 chars). These fall below the 20-char minimum and provide no context for LLM tool selection or hints about when to use this tool vs related alternatives.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 48 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 0 | - | v1 |
Retrieve the score configuration from the project data
Health check endpoint for API status and authentication verification.
Check if the score is currently being generated
List Projects
Update Project
Update Project Data
Update the score configuration in the project data
No tool annotations despite clear risk profiles. deleteProject, deleteScore, and updateProjectData are destructive/write operations but have no destructiveHint annotation. No idempotentHint on read-only tools. Tool annotations are now standard in the MCP spec and help agents reason about safety.
requestBody wrapper adds unnecessary nesting and confusion. In createProject, createScore, updateScore, updateProjectData, the root-level requestBody parameter forces LLMs to nest all fields under requestBody → properties. This deviates from the convention of flattening parameters at the tool root level.
No error handling or recovery guidance. If an operation fails (e.g., project not found, invalid JSON in updateProjectData), there are no hints about what to do next. Errors should be classified as retryable/user-fixable/fatal and include actionable recovery steps.
Parameter descriptions often lack constraint details. Example: 'length_seconds' (video length in seconds 1-600) is good, but many others lack min/max bounds or format specifications. 'The ID of the project' doesn't explain format, length, or example patterns.
Unclear relationships between related tools. createScore, updateScore, deleteScore, getScore, and isScoreGenerating form a cohesive set, but descriptions don't explain prerequisites or sequencing (e.g., 'Call createScore first before updateScore', 'Check isScoreGenerating before calling getScore'). LLMs need explicit dependency hints.
No pagination or result-limiting guidance. listProjects has no limit, offset, or page_size parameters. If the API returns hundreds of projects, all are dumped to the LLM, wasting tokens and degrading reasoning quality. listProjects should accept optional limit and offset/cursor parameters.