Loading ToolBench
Deploy HTML content to EdgeOne Pages with ease. Provide your HTML and let the service handle deployment. Also support to deploy a folder to EdgeOne Pages. Receive a public URL to access your live page.
Two tools with basic schemas and descriptions, but significant gaps in parameter documentation, output schema clarity, and error handling guidance. Both tools accept single string parameters with minimal constraint documentation. No enum constraints, no pagination support, no structured output schema documented. Descriptions are present but generic (avg 95 chars, below 194-char baseline). No error recovery guidance. Tool names follow verb_noun pattern (positive), but lack specificity about what 'deploy' returns or how to handle failures.
Deploy a built frontend directory (or zip file) to EdgeOne Pages. Returns: the deployment URL and project metadata.
Deploy HTML content to EdgeOne Pages, return the public URL
No output schema documented. Tools return plain text responses, but LLMs cannot parse structured data (URL, metadata, status). Agents cannot extract deployment URL or project ID for downstream use.
Parameter descriptions lack format/constraint details. 'value' param says 'Provide the full HTML markup' but does not specify max size, encoding, or what happens if HTML is invalid. 'builtFolderPath' says 'absolute path' but does not clarify: must it exist? Can it be a zip? What if path contains spaces or special chars?
No error recovery guidance. Error handler returns generic 'Error: {message}' with isError=true, but does not categorize errors (retryable vs fatal) or suggest next steps. LLM cannot distinguish 'quota exceeded' (retry later) from 'invalid HTML' (fix and retry).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 53 | 2026-07-28+ | v2 |
Tool names ambiguous about scope. 'deploy_folder_or_zip' combines two resource types in one name, violating single-responsibility principle. Should be 'deploy_folder' and 'deploy_zip' or clarify in description which is primary.
No pagination or result limits documented. If deployment returns large metadata or logs, no indication of max size, truncation, or how to retrieve full results. Risks context window exhaustion.