Short answer
Create one project for one research question. Add the PDFs, correct each source record, and use statuses and tags consistently.
Read in the built-in viewer and attach each important note to the page it describes. Search for known wording when needed, then check Citation History and exported material before writing.
This guide covers CiteFlow. Check requirements and download options before you begin.
On this page
What this workflow solves
A folder can keep PDFs together, but it cannot explain which passage supports which idea. CiteFlow adds a local project library, a built-in PDF reader, source-linked notes, search and citation drafts. It is for reviewing sources you already collected, not discovering every paper on a topic.
Before you start
Write one clear research question and gather the PDFs in a temporary intake folder. A source record stores details such as title, authors, year and DOI. A source-linked note keeps your comment connected to its PDF and page. A status marks progress; a tag groups sources by a meaning you choose.
Use the same note pattern for each source
Start with three papers about one question, such as how students use study breaks. These are practice labels, not research findings. Use one topic tag, such as topic:study-breaks, and a method tag when useful.
- Source and page: which PDF, file position and printed page label?
- Exact quotation: what does the source actually say?
- My note: why does it matter to the question?
- Limit: what does this passage leave unanswered?
Keep your note separate from the quoted words so you can tell them apart later.
Set up a small research project
1. Create a project for one question
Give the project a plain name, such as
Study breaks, and choose the citation style required for your work: APA 7, MLA 9, Chicago 18 Author-Date, Chicago 18 Notes-Bibliography or IEEE.2. Add and identify every PDF
Import one source and let automatic preparation finish. Then check its title, authors, year, publication, and DOI when available. If several sources need attention, use the source review workflow to check them together. You can open the original PDF while reviewing its details. If CiteFlow finds a matching record, reuse it instead of creating a second copy.
A DOI is a published identifier for a source. Automatic preparation can query a DOI or title online and retrieve matching publisher records. Title translation can also run automatically. Entering details manually alone does not prevent these requests. Review Network & data before importing sensitive sources.
3. Use statuses for progress
Open the source actions menu and choose Mark as Unread, Mark as Reading or Mark as Done. These describe reading progress. Use a tag for other labels, such as Excluded; that is not a built-in reading status.
4. Use tags for the comparison question
Open the source record and add the topic or method tags you chose. Reuse the same spelling across the three sources so one tag brings them together. Tags organize your material; they do not rate its quality.
5. Read and write source-linked notes
Open the PDF, choose Note in its toolbar and click beside a relevant passage. Follow the note pattern above, then select Save. Reopen the PDF from the library, navigate to the recorded page and open its note marker to check the link. For an image-only page, use OCR and compare recognized names, numbers and quotations with the page image.
6. Retrieve without losing context
When you remember a word or phrase, use exact search. When you remember the idea, optional Semantic Search BETA can suggest related passages after a local model is installed. Open each result and read its context before using it.
7. Copy, review and export
Copy a passage with citation context and check the saved draft in Citation History. Export only what you need for the next stage. Open the exported file in another program and compare one quotation, page locator and bibliography entry with the original PDF.
Technical note: local search modelsCompare the model choices after you understand the basic search workflow.
Semantic Search BETA supports E5 Large INT8 and BGE-M3 Full Precision. E5 uses less storage and memory. BGE-M3 is the larger multilingual option and needs more memory. Test either model against known passages from your own material. Each model builds its own local project index. The semantic-search guide lists the current download and memory expectations.
Scope and accuracy limits
Problems and limits
- CiteFlow is not a literature discovery database or a formal systematic review platform.
- It does not provide shared cloud collaboration or a full manuscript editor.
- OCR can misread text, and Semantic Search BETA can return a related but unsuitable passage.
- Source preparation and title translation can make automatic online requests using source identifiers or titles.
- Generated citations are drafts that still need source and style-rule checks.
Checkpoint: project ready for writing
- Each source record matches its PDF.
- Statuses describe progress and tags use consistent meanings.
- Each important note returns to a supporting page.
- Search results were read in context.
- Citation History and the export were checked before use.