Adviserry MCP¶
Your newsletter reading, turned into searchable context and finished drafts. Adviserry collects the newsletters and YouTube channels you follow into topic panels, then synthesizes concrete actions from them. Its MCP server puts both layers in front of an assistant: raw source search, the observations Adviserry generated, and the drafted deliverables themselves.
Server type: Remote (Streamable HTTP)
Auth: Bearer token from Settings, MCP Server Access (no OAuth 2.1, so ChatGPT custom connectors are not supported yet)
Endpoint: https://adviserry.com/api/mcp
Tools: 13
Rate limit: 500 tool calls per rolling hour, per user
Category: Research
Built by: adviserry.com
What it exposes¶
The server publishes thirteen tools. Source search: search_newsletters (query, optional panel filter, limit up to 10) and search_context for PDFs, Word docs, spreadsheets, presentations and web pages you uploaded as context. Panel and issue reads: list_panels, get_recent_issues (days up to 30), get_newsletter_summary and get_topics.
Synthesis reads: get_insights returns the AI-generated connections, trends and action items across sources, filterable by unread, read, saved or all. The action layer is where it differs from a search tool: search_actions finds drafted deliverables by topic against each action's title, its "why now" rationale and the full draft, filterable by panel, by domain (project, investment, personal, learning), by effort (quick, medium, deep) and by status; get_actions lists the current batch newest-first; get_action returns one in full with every source it was built from, author and date included. update_action_status marks an action shipped, read, dismissed or back to unread, and shipped actions teach future synthesis.
Context writes: get_user_context returns the advisor profile (role, current projects, challenges, goals, freeform notes) and update_context changes it. Updates default to append because the profile spans more than one project; anything that would remove existing text returns a preview and saves nothing until you pass back the confirm token.
Connecting¶
Copy your MCP token from Settings, MCP Server Access, then add the server to your client config with the endpoint and a bearer header. Claude Desktop, Claude Code, Cursor, Gemini CLI and OpenClaw are supported over the same HTTP endpoint. Two setup details matter: the endpoint must be the bare host adviserry.com because the www. host returns a 308 redirect most MCP clients will not follow, and the mobile app asks for your Adviserry user ID in a second field, which is not a credential, since requests are authenticated by the bearer token alone.
Notes¶
The rate limit returns HTTP 200 with a JSON-RPC error inside the body (code -32000) rather than a 429, so a client that only inspects status codes will miss it. Regenerating a token invalidates the old one immediately, and clients read their config once at launch, so replace the token and fully restart the client rather than reloading. ChatGPT is not supported because its custom connectors require OAuth 2.1 with dynamic client registration.
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.