MCP server for Jira and Confluence with search, view, create, and edit capabilities. Converts Markdown to ADF for beautiful native formatting.
15 tools with consistently strong naming (verb_noun format), clear descriptions (mostly 100-200 chars), and properly structured JSON Schema. All tools declare required parameters. Parameters have type definitions and descriptions. However, output schemas are NOT documented, LLMs cannot know what fields to expect from responses. Error handling and recovery guidance are absent. Some parameter descriptions lack constraints (e.g., limit ranges not enforced in schema). Tool composition is good (separate search/get/create/update operations), but lacks pagination guidance and chaining hints.
Add an inline comment to an existing Confluence page. REQUIRED: Always provide Markdown content. Supports bold, italic, links, and basic formatting. Content is automatically converted to Confluence storage format (HTML).
Check whether the configured API token is valid and has the required permissions. Validates authentication, read/search access, and optionally write access to a specific page.
Create a new Confluence page in a given space. REQUIRED: Always provide Markdown content. Supports headers, lists, bold, italic, code blocks, tables, links, images, and more. Content is automatically converted to Confluence storage format (HTML). Optionally nest the page under a parent page.
Read a Confluence page by its ID. Returns the page title, space, version, URL, and full body content in Confluence storage format (HTML).
Get the version history of a Confluence page. Returns a list of versions with author, timestamp, and optional version message.
No output schema documentation. LLMs cannot know what fields to expect from responses (e.g., search_confluence returns what fields? jira_get_issue returns what structure?). This forces LLMs to guess and risk parsing failures on multi-step workflows.
No pagination guidance. search_confluence and jira_search accept 'limit' but don't document if results can exceed limit, if a next_cursor/startAt is returned, or how to chain calls. Agents may miss results or retry inefficiently.
Parameter constraints not enforced in schema. limit parameters state '1-100, default 25' in description only; JSON Schema lacks minItems/maxItems or enum constraints. LLMs may pass invalid values (0, 1000) without validation.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 65 | 2026-07-28+ | v2 |
Add a comment to a Jira issue. Comment content is Markdown-formatted and automatically converted to Atlassian Document Format (ADF).
Create a new Jira issue (Story, Bug, Task, etc.) in a given project. New issues are created unassigned by default. REQUIRED: Always provide a markdown-formatted description.
Get full details of a single Jira issue by its key (e.g. 'PROJ-123') or numeric ID. Returns summary, status, type, priority, assignee, reporter, description, labels, and timestamps.
Get available transitions for a Jira issue. Returns a list of transition names that can be performed on the issue in its current state.
Search Jira issues using a JQL (Jira Query Language) query string. Returns a list of matching issues with key details including status, assignee, priority, and description.
Transition a Jira issue to a new status (e.g. 'In Progress' → 'Done'). Returns the updated issue key and new status.
Update an existing comment on a Jira issue. Comment content is Markdown-formatted and automatically converted to ADF.
Update an existing Jira issue. Supports updating summary, description, status, assignee, priority, labels, and custom fields. Description is Markdown-formatted and automatically converted to ADF.
Search Confluence pages using a CQL (Confluence Query Language) query string.
Update an existing Confluence page. REQUIRED: Always provide Markdown content. Supports headers, lists, bold, italic, code blocks, tables, links, images, and more. Content is automatically converted to Confluence storage format (HTML). Automatically handles version bumping.
No error recovery guidance. If search_confluence returns 0 results, if create_jira_issue fails due to invalid status, or if permissions are insufficient, tools do not suggest what the LLM should do next (try different search terms, check issue type availability, verify credentials).
No tool annotations. Tools like create_confluence_page, update_confluence_page, jira_create_issue, jira_update_issue, jira_transition_issue lack destructiveHint or idempotentHint annotations. LLMs cannot determine retry safety or whether side effects are permanent.
Markdown conversion responsibility unclear. create_confluence_page, update_confluence_page, and jira_create_issue accept Markdown and claim automatic conversion to storage format (HTML/ADF). Descriptions say 'automatically converted' but do not specify what Markdown subset is supported, what happens to unsupported syntax, or what the LLM receives back (HTML preview? Confirmation?). If conversion fails, no error guidance.
Chaining dependencies not documented. If an LLM calls search_confluence, then wants to update the page, it needs pageId from the search response. No description says 'Call search_confluence first, then extract pageId for get_confluence_page.' This forces agents to plan multi-step workflows without guidance.
Parameter overload in update tools. jira_update_issue accepts optional summary, description, status, assignee, priority, labels, all optional, all modifiable in one call. No description clarifies: are all fields updated together, or are unspecified fields left alone? Can an LLM pass only 'status' and leave other fields unchanged, or is a full object required? Ambiguity leads to silent partial updates.
Idempotency not stated. If an LLM retries add_confluence_comment or jira_add_comment due to a transient network error, will it create a duplicate comment? Descriptions do not guarantee idempotency, forcing agents to assume worst case and avoid retries.