Refine.ink
Refine is an AI-powered proofreading and critique engine built for researchers, academics, and technical writers — the people whose work is judged on the soundness of its arguments, not just its grammar. Unlike generic writing assistants that draft or paraphrase text for you, Refine deliberately does the opposite: it acts as a rigorous expert reviewer. You upload a document (PDF, DOCX, and more), and Refine analyzes it the way a demanding peer reviewer or editor would — interrogating the logic of your arguments, auditing the statistics and mathematics behind your results, flagging unsupported claims, methodological gaps, inconsistencies, and weaknesses in structure and exposition. It returns detailed, anchored feedback comments tied to specific passages, each with a clear justification and concrete suggestions for improvement. The core philosophy is critique, not draft. Refine never hands you replacement prose to paste in; it tells you what's wrong and why, leaving the authorship — and the thinking — with you. This is the wedge that distinguishes it from tools like ChatGPT, Thesify, or coarse.ink: it strengthens the author rather than substituting for them, which is exactly what scholarly and grant-funded writing demands. Beyond one-shot feedback, Refine lets you chat with an AI assistant about each comment — to get explanations, push back, or work through how to address an issue. It tracks which comments you've addressed across revisions, supports sharing reviews with collaborators, and is increasingly used by institutions: universities, research organizations, and partners like the Econometric Society value its detailed technical auditing of results and arguments. Refine serves individual researchers (per-document and subscription plans) as well as teams and institutions, with a growing focus on grant applications and high-stakes academic submissions where the cost of a weak argument is highest.
- Integration type
- Connector
- Verification status
Community connector- Platform
- Claude
- Primary Subcategory
- Academic Writing & Citation Tools
- Secondary Subcategories
- None listed
- Brand
- Refine
- Access
- Account required
- First tracked
- 2026-06-24
- Tool count
- 35
- Geography
- US
The Primary Subcategory used for this profile’s headline score.
Other Subcategories where the Integration is visible.
Claude Discoverability Score
Community connectors cannot be organically discovered yet
Refine.ink is a Community connector, so Claude cannot surface it organically yet. It becomes eligible for organic discovery after Claude verifies it. Learn how Anthropic reviews Connector Directory submissions in the official Connector Directory guide.
Get notified when this connector becomes Verified
Enter your work email and we’ll notify you when it can be organically discovered in Claude.
No spam. Unsubscribe any time.
Competing in Claude Academic Writing & Citation Tools
View CategoryHow the Discoverability Score works
The Claude Discoverability Score will be the percentage of contested conversations where Refine.ink appears in the connector picker or Claude invokes it directly. Found and Positioned will be shown as diagnostics, not score inputs.
Community connectors appear on this scale after Claude verifies them and they become eligible for organic discovery.
FoundDiagnostic
Whether Claude found your Connector in connector search. It must be Found before it can reach the picker, but the score counts picker appearances—not search results.
PickedMain score
How often your Connector appeared in the picker, or Claude invoked it directly, across contested conversations. This percentage is the Discoverability Score; the headline number is rounded.
PositionedDiagnostic
What position your Connector appeared in when it was shown in the picker. This shows prominence, but it does not affect the score.
35 tools agents can invoke
How do I improve a Community connector's discoverability?
The levers are the listing surface agents actually read: names, descriptions, keywords, tool metadata, and registry health. Which lever matters depends on where discovery breaks, which is what continuous measurement shows.
Where is this profile measured?
This profile uses the geography attached to the latest public registry snapshot: US. Locale tags are intentionally omitted.