BlueCLInput and model processing
Submitted text is sent to the configured model providers for summarization and task extraction.
Mindflow MCP / Knowledge workflows
From a note to a clear next step.
Mindflow turns study notes into organized learning material and actionable tasks, then saves the result to Notion. Independent summary and extraction workers run in parallel to keep the workflow moving.
Development complete
Three steps
From the first step to a useful result.
- 01
Bring your source
Provide text or a supported file through the connected client.
- 02
Organize the note
Summaries and task/date extraction run in parallel.
- 03
Review the result
Read the summary and extracted actions before using the connected Notion workflow.
PRODUCT / IN VIEW
Connected tools.
A project illustration of AI tools and workspaces. Select the image to view the original.
Getting started
Start with the right setup.
Access & setup
- Read the repository and choose an MCP or API client.
- Configure provider credentials on the server and connect the Notion integration if you use publishing.
- Start with a short note, inspect the summary and tasks, then connect your own workflow.
Permissions & data
Understand what the workflow uses.
Notion connection
The Notion workflow writes through the configured integration. Grant access only to the workspace pages you intend to use.
This is product usage guidance. The BlueCL website privacy notice covers this company website; consult the service’s own documentation for its deployed data handling. Website privacy ↗
Behind the product
How the product works.
A useful learning workflow needs more than a short summary. Learners must identify concepts, find assignments and deadlines, and move the results into a workspace. Repeating those steps by hand interrupts the work itself.
01Text or fileAuthenticated FastAPI input
02Parallel interpretationSummary + task/date extraction
03Page creationNotion page from extracted fields
04Connected outputAppend study content; return page URL
Model integration
The Python backend uses Claude Haiku 4.5 for study-material generation through the Messages API. The composite tool runs it alongside GPT-4o mini task/date extraction; individual MCP tools also expose summarization and schedule extraction.
Implementation details
The fast-save implementation creates the Notion page when extraction finishes, then appends the Claude summary when it is ready. Notion blocks are split into batches, with bounded retry handling for rate limits and server errors. The web API exposes /process and /process-file.
Further improvements
Develop evaluation cases for Korean summaries and task/date extraction, improve partial-failure recovery, and measure the latency and cost of each parallel branch.
When you need a hand
Frequently asked questions.
Does it work offline?
Model API calls and Notion integration require network access.
Where is the result saved?
Publishing creates a page in the connected Notion workspace. Review the generated material and extracted dates there.
Can I use only part of the workflow?
The MCP implementation exposes individual tools as well as a composite workflow. See the source for the available interfaces.
Product and source information updated: October 8, 2026