Loading ToolBench
Read-only MCP server for deep Jira Data Center 10 introspection — 76 tools for admins
Server has 76 tools with consistent naming (verb_noun pattern) and descriptions present for all sampled tools. However, critical gaps exist: input schemas are visible only for 6 tools in the provided code; parameter descriptions are minimal or absent for most tools; output schemas are undocumented; error handling lacks recovery guidance. The 'analyze_project_config_chain' and 'search_config' tools show good description depth (100+ chars), but most tools lack parameter-level documentation. No evidence of pagination, result limits, or structured output documentation across the 76-tool suite.
Resolve the FULL scheme chain for a project and report inconsistencies. Shows: project → issue types → workflow scheme → workflows, issue type screen scheme → screen schemes → screens, field configuration scheme → field configs. Reports issues (missing mappings) and warnings.
Resolve effective permissions on a project by walking groups, roles, and grants. Pass username to get every permission a user holds, and/or permission to get every user that holds it. At least one is required.
List application links to Confluence, Bitbucket, Bamboo, etc., with type, URLs, and authentication status.
List all registered event listeners (built-in, plugin, ScriptRunner). Listeners cause side effects invisible to the REST API.
List all Jira scheduled services (mail handlers, backup services, etc.) with their cron schedules. Sensitive properties are redacted.
Output schemas undocumented. No evidence of return type definitions, pagination support, or result limits across the 76-tool suite. LLMs cannot plan downstream calls or extract required fields.
Parameter descriptions missing or minimal. 'get_effective_permissions' has 3 params but only 2 have descriptions; 'project_key' lacks format guidance (e.g., '2-10 uppercase letters'). Ambiguous params force LLM guessing.
No pagination or result limits documented. 'list_listeners', 'list_scheduled_services', 'list_application_links' return unbounded lists; no mention of page/offset/limit params or total counts. Large results risk context window exhaustion.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 61 | 2026-07-28+ | v2 |
Free-text search across all config entities: fields, screens, workflows, workflow schemes, issue type schemes, permission schemes. Case-insensitive search on names and IDs.
Error handling lacks recovery guidance. Tool handler catches errors and returns JSON with 'error' and 'tool' fields, but no actionable next steps (e.g., 'Try search_users() first' or 'Retryable: yes/no'). Agents cannot self-correct.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All 76 tools are read-only but this is not declared in the schema. Clients cannot optimize caching or retry logic without explicit hints.