Connects AI agents such as Claude, Cursor, ChatGPT, and Gemini to documentation published with ClickHelp, a help authoring tool. Agents can search a documentation portal, navigate its table of contents, and read topics, so answers come from the current published docs instead of guesses or a stale copy. Read-only access to public publications, no authentication required. Every ClickHelp portal has its own endpoint at /published-docs-mcp-server; this listing points to the ClickHelp user guide.
Scored across 11 tools
Most tools are clearly separated: search_portal does full-text search, find_toc_nodes searches titles, get_toc_node/get_child_toc_nodes/get_toc_tree handle TOC structure, and list_projects/list_topics/get_topic handle content. The four TOC-related tools (find_toc_nodes, get_child_toc_nodes, get_toc_node, get_toc_tree) overlap somewhat, but their descriptions do a good job of spelling out when to use which.
The naming follows a consistent verb_noun pattern (get_project, list_topics, find_toc_nodes, search_portal, ask_ai) with clear snake_case throughout. system_connection_check is the only name that doesn't match the verb_noun pattern, but its use as a diagnostic makes this acceptable.
With 11 tools, the count is well within the optimal range of 3-15. Each tool has a distinct role: discovery, TOC traversal, content retrieval, and search all earn their place, and the count maps cleanly to the portal's read-only documentation scope.
The tools cover the read-only surface completely: project and publication listing, TOC traversal, topic listing and full-content retrieval, full-text search, and an AI Q&A endpoint. There are no write operations, but that seems consistent with the 'Published Docs' read-only scope, so the gaps are minor.
Asks the portal's AI assistant a question and gets an answer based on the portal's documentation content
Inputs: query (required), chatId, projectId.
Search TOC nodes by title substring or topicId glob pattern (e.g. 'install-*'), returns plain-text search results info and YAML-formatted TOC nodes, if found. Use this when you know part of a node's name but not its nodeId. For full-text content search, use search_portal instead. TOC node IDs and topic IDs are unique within a project/publication scope. Works for both projects and publications.
Inputs: projectId (required), maxResults, isShowInToc, titleContains, +1 more.
Returns flat list of children for a TOC node. TOC node IDs are unique within a project/publication scope. Works for both projects and publications (read-only).
Inputs: nodeId (required), projectId (required), isRecursive.
Gets detailed information about a specific project or publication. Project IDs are unique across the entire portal. Returns id, url (full URL with protocol and domain), parentId (null for projects, project ID for publications), title (name), visibility (Private/Public/Restricted), topicCount (number of topics), modifiedOn (ISO timestamp of the most recent topic modification, or the creation timestamp if there are no topics), type (Project/Publication), lang (language code, e.g. 'en-US'), and baseLangProjectId (base language project ID for translations). For projects, publications lists the publications made from the project. Translation projects have limited topic editing: only workflow fields (status, assignee, owner) can be changed on their topics. Publications are auto-generated read-only project copies - their content cannot be modified and any changes would be lost when the publication is regenerated.
Inputs: projectId (required).
Returns single TOC node with parent/sibling context for understanding node position. TOC node IDs are unique within a project/publication scope but may duplicate across different projects. TOC nodes in translation projects and auto-topic TOC nodes are completely immutable - they cannot be updated, moved, reordered, or have children added to them. Publication TOC nodes are read-only too (publications are auto-generated read-only project copies).
Inputs: nodeId (required), projectId (required).
Returns hierarchical TOC tree as YAML. Use this to explore project structure, find nodeIds, or understand parent-child relationships. For large projects, use parentId to scope to a subtree and maxDepth to limit depth. Returns: title, topicId, nodeId, visibility for each node. Note: TOC node IDs are unique within a project/publication but may duplicate across different projects. Publication TOC is read-only (publications are auto-generated project copies).
Inputs: projectId (required), maxDepth, parentId, includeHidden.
Get a single topic from the portal with full content. Returns topic metadata, body source and compiled content. The 'body' field contains source HTML (editable, may contain custom tags like ch:var/ that should be typically preserved unless surrounding/related content is removed), while 'md'/'html' is the compiled readable output. Use format='md' for readable content, format='html' for raw HTML output. MUST use isReturnBody=true and use body as topic content if the retrieved data is supposed to be written back to the portal to prevent HTML formatting loss. NEVER use compiled content from the md/html fields as topic content for editing to prevent data loss. Note: Auto-topics (isAutoTopic=true) are read-only and cannot be updated, deleted or used as new parents for TOC update operations. Translation project topics can only have workflow fields (status, assignee, owner) updated. Publication topics are read-only (publications are auto-generated project copies).
Inputs: topicId (required), projectId (required), format, isReturnBody.
Lists all projects and publications in the portal available to the current user. Project IDs are unique across the entire portal. Returns id, url (full URL with protocol and domain), parentId (null for projects, project ID for publications), title (name), visibility (Private/Public/Restricted), modifiedOn (ISO timestamp, publications only), type (Project/Publication), lang (language code, e.g. 'en-US'), and baseLangProjectId (base language project ID for translations). Translation projects (those with baseLangProjectId set) have limited topic editing: only workflow fields can be changed. Publications are auto-generated read-only project copies - any changes would be lost when the publication is regenerated. Note that modifiedOn is populated for publications only, use get_project to get that info for projects.
Inputs: no named parameters.
List all topics in a project or publication. Returns metadata only (id, title, status, dates, isAutoTopic) without content for performance. If you need topic URLs, you can construct them as https://{portalDomain}/articles/{project-id}/{article-id}, e.g. https://your-portal.example.com/articles/my-project/my-topic. Use get_topic to fetch full content of a specific topic. Note: Topics marked with isAutoTopic=true are read-only and cannot be updated, deleted or used as new parents for TOC update operations. Topic IDs are unique within a project/publication scope but may duplicate across different projects. Publication topics are read-only (publications are auto-generated project copies that may be regenerated at any time).
Inputs: projectId (required), pageSize, pageNumber.
Performs a full-text search across the portal to find topics containing the specified keywords
Inputs: query (required), count, language, projectIds.
A connection diagnostic tool that returns the provided message and does not affect anything. Do not use this tool unless explicitly asked by the user to perform a connection test.
Inputs: message (required).
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Add one secure layer between your agents and this server.
