Allows managing a Notion database of automation requests, including listing pending requests, saving priority scores with rationale, and retrieving ranked backlogs.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Request Triage MCP Server list pending requests"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Most of the "AI agent" projects I'd built before this were really just RAG pipelines with extra steps; just ask a question, retrieve some context, get an answer back. Nothing was actually deciding anything. This one does: give it a pile of automation requests from different teams and it works out, on its own, which ones matter most and why.
Feed it a backlog of requests sitting in a Notion database eg "Sales wants X," "Support wants Y," each tagged with a rough urgency and effort. Point the agent at it, and it scores every request, writes a short rationale for each score, and hands back a ranked list. Nobody told it the order of operations, it figures out by itself.
Related MCP server: sprinter
Three tools, exposed through an MCP server:
list_pending_requests — pull everything still waiting to be triaged
save_priority_score — write a score (1-100) and a one-line rationale back to a request
get_ranked_backlog — pull the final ranked list once everything's scored
Claude gets the goal and those three tools, nothing else. It calls whichever one it needs, looks at what comes back, and decides the next move. This loop is the whole project. agent.py is basically that loop and nothing more.
Notion is just the system of record here, not the point. Swap it for a database or a Slack channel and the agent logic doesn't change. This is the actual value of building it on MCP instead of writing one-off glue code to Notion's API directly in the agent.
Claude (Sonnet) for the reasoning
MCP (mcp Python SDK) for the tool layer; server in mcp_server.py, client/loop in agent.py
Notion API for storage (raw REST calls in notion_helper.py, no SDK; wanted to see the actual request/response shape)
pip install -r requirements.txt
cp .env.example .env # fill in your Anthropic key + Notion integration token + database ID
python notion_helper.py # seeds a handful of sample requests
python agent.py # watch it triage themYou'll need a Notion integration with access to a database that has these columns: Request (title), Team (select), Urgency (select), Effort (select), Status (select: Pending/Scored), Priority Score (number), Rationale (text).
I wanted to actually build something that takes actions instead of just answering questions, and prioritization-under-competing-demands felt like the most honest test of that; it's a real problem and "here's my reasoning for the ranking" is a much better answer than a black-box score.
This server cannot be deployed
Agent-complete, permission-scoped product operations for Priorify workspaces.
Capture feature requests and bug reports from chat into a searchable, AI-categorized backlog.
AI-native task management: list, create, update and archive tasks with rich context for AI agents
Turns vague automation requests into tool stacks, prompts, QA checks, and human boundaries.
