This tool does not answer in place of the reader. Its function is to take a free-form question, then open a path through the atlas: where to begin, which path to read, which evidence to review, and which point of tension or scrutiny should not be skipped. It may also display the Qur’anic Mirror of the Shahrur Atlas when the question is close to Qur’anic loci in the atlas.
The intended format is simple:
Start from this concept, then read this path, then review these pieces of evidence, then pay attention to this tension.
Tool status
This tool’s share has temporarily ended for research purposes. Come back later; the share may open again, or sign up for the newsletter to receive atlas updates.
This tool is part of the Muhammad Shahrur Atlas project: it helps suggest a reading path within the material, but it is not the basis of the project nor a substitute for reading the published pages, sources, and evidence.
How does it work?
The generator reuses atlas pages and internal links. It does not create a new database, nor does it replace atlas pages with an automated summary. When the question arrives, it searches for pages close from four angles:
- A concept or entry point from which the reader can begin.
- A published reading path close to the question.
- Evidence that can be checked: pages, concepts, internal links, Qur’anic loci, or sources.
- A point of critique or tension that helps the reader avoid turning the result into a final answer.
After that, the model arranges these results into a single reading path, with links to the sources on which it relied.
If the RAG service returns Qur’anic Mirror fields, they appear as an additional scrutiny layer: the closest concept, the question’s locus in the atlas, nearby verses or verses used in Shahrur’s arguments, and what cannot be inferred from the retrieved material. This layer does not interpret the Qur’an automatically; rather, it points to semantic proximity, a review path, and evidence that requires reading.
What is the difference between it and an automated answer?
An automated answer tends to close the question. A reading path, however, leaves the question open to the material: read the page, check the evidence, compare the loci, then pay attention to the objection or tension.
That is why a general statement like “Shahrur sees freedom as the basis of the human being” is not enough. The better result tells the reader: start from the concept of freedom, then read the path of human being, freedom, and responsibility, then review the evidence for no compulsion in religion and civil state, and then pay attention to the tension involved in transferring the concept from ethics to law.
Limits of the result
If the search does not find sufficient evidence, that must appear in the result. Weak coverage is an editorial signal that the atlas may need a clearer path, piece of evidence, scrutiny page, or page.
The generator remains part of the reading tools in semantic search and retrieval and the Qur’anic Mirror, not a substitute for them. Search presents nearby passages, the answer summarizes from the passages, while this generator turns the question into a reading path within the atlas.