Changelog
Every correction to a hadith page, and every change to the method or the rubric, is listed here with its date. To report a problem, use the link at the bottom of any hadith page or open an issue.
16 November 2026: versions of the same report (step 5)
- Each hadith page now lists the other versions of the same report found in the six books, with their quality levels and how each was matched. 22,991 hadith are in 6,564 groups. Reports that share wording but are not the same report, such as the same saying from a different Companion, are listed separately as "similar wording".
- The grouping was rebuilt. The first version, used to rank the curation queue, sometimes chained unrelated reports together. For example, the report of 'A'isha's age at marriage was grouped with reports about garlanding sacrificial animals, because a narrator chain left inside one text made them look alike. Chain words are now removed before comparing, a single weak link can no longer join two groups, and the compiler's own cross-references ("with this chain, the same") are used.
- Search has a new filter, "Versions": no other version found, or versions in 1 to 6 books.
- A hadith now also carries the topics of the chapters where the compilers placed its other versions.
- The curation queue is unchanged: it keeps the grouping it was examined with.
- The method is on the Method page. The groups were made by a script and have not been reviewed by a person.
16 November 2026: the Qur'an, and links between hadith and verses
- The Qur'an is now on the site: all 114 surahs, with the Arabic text from the Tanzil Project and the Saheeh International English translation. Each verse lists the hadith linked to it, with their quality levels.
- Every hadith now shows its link to the Qur'an: it explains a verse, quotes a verse, or has no link recorded. Explaining means the compiler placed it in a Qur'an-commentary chapter, it reports an occasion of revelation, or it asks about or comments on a verse. The rules are fixed and listed on the Method page, and each quotation shows the matched words.
- Search and Browse have a new filter, "Link to the Qur'an".
- Links were found by a script and have not been reviewed by a person. "No link recorded" is not a finding that a hadith has no bearing on the Qur'an.
16 November 2026: new design
The site has been redesigned to be easier to read. The rubric, the levels and the concerns are unchanged.
- On a hadith page the text now comes first. On wide screens a sidebar shows the quality level, the reasons for it, the status of each approach and the references. On phones the level is shown under the heading.
- Every hadith in lists and search results has a square marker in its level's colour, always beside a tag with the level's name. On a hadith page the level is a solid block of its colour with the name in large type.
- Type is set in IBM Plex throughout (sans, serif for English text, mono for reference data, sans Arabic for labels), with Amiri kept for hadith text.
- The home page explains the four levels with their counts and links to each level. Each book shows its spread of levels.
- The menu is shorter. People, Curation and the Changelog are now linked in the footer.
16 November 2026: rubric version 3, and browsing by topic and level
These changes follow a review that found the site too reluctant to record concerns about reports in Sahih al-Bukhari and Sahih Muslim. Their consensus status in the Sunni lens never protected them in the rubric, but the way findings were classified often did.
- Historical-critical dating rule revised. A study that dates a report, or a substantive clause in some versions, after the Prophet's lifetime now counts as substantive. This includes a clause the author judges was added or inserted later. A new rule covers studies that find a Successor or later common link but argue for an earlier origin on other grounds: these count as minor, with the author's own dating stated. Six findings were reclassified under these rules:
- Goldziher on "actions are by intentions": transmitted only in Medina in the early period. Now substantive.
- Kister on the added stipulations in "I have been ordered to fight the people": substantive, for the expanded versions only.
- Eich, reporting van Ess, on the later clause about deeds in the hadith of creation in the womb: substantive.
- Kesgin on the clause "evil omen is in three things": judged an insertion, so substantive for the three versions carrying it.
- Faghhizadeh and Dastranj, who find al-Zuhri as the common link of the seven ahruf reports: minor.
- Akpınar, who finds Nafiʿ as the common link of the zakat al-fitr report: minor.
- Reports not of the Prophet are capped at Good quality. A report giving the words or practice of a Companion (mawquf) or a later figure (maqtu') can no longer be High quality. 801 reports are affected. 150 of them carry a grader's own maqtu' label. The other 651 were found by reading every report whose Arabic never names the Prophet: 1,426 in all. Each page quotes the deciding words, and says the reading was done by an AI model and not yet reviewed by a person.
- Assessments of the person a report starts from. 31 cited assessments of 12 people at the head of chains are now recorded and applied to every report that starts from them. The sources are Shia works already in the source set, classical Sunni works (Ibn Hajar, al-Nawawi), and historical-critical studies (Goldziher 1884 and 1890, Sprenger 1865). The severity rule is fixed: rejection as a narrator is substantive; criticism short of that is minor. All of them are on the new People page, with published defences linked. Assessments of Successors ('Urwa, al-Zuhri, 'Ikrima, Hisham b. 'Urwa) and of Ka'b al-Ahbar are recorded there but not applied, because the reports that start from them cannot yet be read reliably from the chains.
- Effect on levels in the two Sahihs:
- Sahih al-Bukhari: High quality 7,294 → 4,269; Good 256 → 1,740; Some issues 30 → 1,571.
- Sahih Muslim: High quality 7,105 → 5,174; Good 235 → 1,062; Some issues 20 → 1,124.
- New: Browse. Every quality level and every topic has its own page, with counts by book. Any count opens one list of those hadith from all six books together. The search page can now filter by topic and by whose words a report gives, with or without search words. Topics come from each book's chapter headings by a fixed mapping (Method).
16 November 2026: step 4 completed for the queue of 500 reports
- Hady al-Sari, all 110 entries published. The 16 held earlier were settled. A new rule covers entries where Ibn Hajar records an objection without ruling on it: these count as minor (see Method).
- Classical critiques of Sahih Muslim: 118 objections, from al-Daraqutni's al-Tatabbu' and Ibn 'Ammar al-Shahid's 'Ilal. Each is shown with al-Nawawi's answer from his commentary on Muslim, where he gives one. 22 more are held, with reasons: unclear positions, objections that turn out to uphold the report, duplicates, or reports missing from our edition.
- al-Tirmidhi's own verdicts ("hasan sahih", "gharib" and so on) are now read from his text and listed as grades for 2,036 hadith in his Jami'. Verdicts phrased in other ways are not read yet.
- Grade reading corrected: "Hasan Gharib Sahih" now counts as authentic, like "Hasan Sahih Gharib". Before, the two orders were read differently.
- Shia lens: all 500 queue reports examined against a fixed set of ten works (shia-v1), listed on the Curation page. This produced 18 concerns: 3 serious, 4 substantive and 11 minor. The other reports say "examined: no specific critique found", with the works searched.
- Historical-critical lens: 31 published studies opened and read, giving 13 concerns (3 serious, 3 substantive, 7 minor) and 9 studies that argue for an early origin. 476 of the 500 reports are examined. 22 known studies could not be opened (for example Schacht's Origins, which has no openly licensed copy). The 363 hadith they discuss stay "not yet examined", and their pages name the study.
- Rubric (version 2): a lens now also counts as examined when the hadith was checked against that lens's listed source set. Such a check adds no concern, so no level changes from it alone. Fixed severity rules for Shia and historical-critical concerns were added to the Method page.
- Page numbers corrected. OpenITI page markers close the page they name. Earlier citations read them as opening it, so many pages were one too low. Every page is now read from the source text.
- Each concern from these sources says it was checked by an AI model reading the source and has not yet been reviewed by a person.
16 November 2026: classical critiques and the curation queue (stage 4, first part)
- 94 classical critiques of Sahih al-Bukhari added as Sunni-lens concerns, applied to 232 hadith in Bukhari and Muslim. They come from chapter 8 of Ibn Hajar's Hady al-Sari, which covers al-Daraqutni (in al-Tatabbu'), Abu Mas'ud al-Dimashqi, Abu 'Ali al-Ghassani, al-Khatib, 'Abd al-Ghani b. Sa'id and al-Isma'ili. Each records Ibn Hajar's reply as its published response, and each quotes both word for word.
- 16 more entries from the chapter are held back, unpublished: Ibn Hajar's position is not stated in the entry, or the matching hadith is in doubt. The reasons are on the Curation page and in the worksheet.
- New severity rule for classical critiques, added to the Method page. Severity follows how the defending scholar treats the objection: rejected, or a defect in one route only, counts as minor; granted against the report counts as substantive. The rubric itself is unchanged.
- Level changes under the unchanged rubric:
- 225 hadith in Bukhari and Muslim move from High quality to Good quality (a minor concern).
- 7 move from High quality to Some issues (Ibn Hajar grants the objection).
- Verbatim quotes are checked against the source texts by the build and the tests. A quote not found in its source stops publication.
- Each concern says who checked it. These were checked by an AI model reading the sources and have not yet been reviewed by a person.
- New Curation page with the queue of the 500 most widely transmitted reports. Versions are grouped automatically by shared wording; the grouping is approximate and will be refined in stage 5.
16 November 2026: chains of transmission (stage 3)
- Each hadith page shows its chain of transmission, split from the text automatically, with each narrator matched to Ibn Hajar's Taqrib al-Tahdhib and, where the lineage agrees, al-Tusi's Rijal. Labelled "automatic, not yet verified".
- New narrator pages quote the full Taqrib entry and list the chains where the narrator was matched.
- The chains of all versions covered by a concern are drawn together in one diagram (first: the 20 versions of the marital-age hadith).
- Concern records now name the method the cited study used (Little 2022: isnād-cum-matn analysis).
- No change to the rubric or to any level: automatic narrator matches are shown but not counted until checked.
16 November 2026: quality levels (stage 2)
- Rubric introduced (version 1), as code with tests and published on the Method page. Every hadith now has a level, marked provisional while any lens is unexamined.
- Grades from the source edition are read into categories (authentic, hasan, weak, very weak, fabricated); each hadith page shows the category beside each grade.
- Concern records added, with full citations and a verification note. First record: Little (2022) on the hadith of ʿĀʾišah's marital age, applied to its 20 versions in the six books (historical-critical lens, substantive).
- Modern scholarly (historical-critical) concerns are highlighted at the top of each hadith page, with their effect on the level, and listed on a new Scholarship page.
- Search can filter by quality level and concern type.
15 November 2026: first release (stage 1)
- All six collections in Arabic and English from hadith-api, browsable by book and chapter.
- Keyword search in Arabic and English. Arabic search ignores vowel marks and normalises alif, ta marbuta and ya forms.
- Grades recorded in the source edition are listed on each hadith page, with the grader's name.
- Each hadith shows the Shia and historical-critical approaches as "not yet examined".
- Quality levels are described on the Method page but not yet assigned; the rubric arrives in stage 2.