Loading ToolBench
MCP server for Jira integration that allows creating and updating Jira issues
Two tools with basic schemas and minimal descriptions. Both tools have input schemas with typed parameters, but descriptions are generic (20-30 chars), lack actionable context, and omit error guidance. No output schemas documented. Parameter descriptions are present but sparse. Missing enum constraints for 'issueType', 'priority', 'status'. No error handling guidance or recovery paths. Tool composition is reasonable (separate create/update), but descriptions do not explain when to use each or what they return.
Create a new Jira issue
Update an existing Jira issue
Tool descriptions are too short (20-30 chars) and lack actionable context. 'Create a new Jira issue' does not explain what fields are required, what the tool returns, or when to use it instead of update. LLMs cannot reliably select the tool without richer guidance.
No output schema documented. Callers cannot know what fields the response contains (issue key, URL, status, etc.). This forces LLMs to guess and breaks tool chaining.
Parameters 'issueType', 'priority', 'status' accept free-form strings with no enum constraints. LLMs will hallucinate invalid values like 'Critical' instead of 'High', or 'In Progress' instead of 'In_Progress'. Jira APIs are strict about these enums.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 42 | 2026-07-28+ | v2 |
No error handling guidance. Code returns generic error messages like 'Failed to create JIRA issue: <error>' without telling the LLM what to do next (retry, ask user, check credentials, etc.). Agents cannot recover.
Parameter descriptions are minimal. 'Labels to attach to the issue' does not explain format (array of strings), max count, or character limits. 'Custom fields for the issue' is vague, LLMs cannot infer which custom fields exist or how to structure them.