About ELGODS

ELGODS is a free, open spiritual encyclopedia covering deities, sacred texts, practices, festivals, and religions across 351+ tradition tags worldwide, with a rooted focus on the traditions of Bhārata. Our mission is to make wisdom from every tradition equally accessible — without paywalls, without monetization, forever.

Each entry is structured for three audiences — beginner, intermediate, and scholar — so a curious reader and a researcher can both find their level. Where possible we cite primary sources and traditional commentaries.

Religious knowledge carries real weight for living communities. We prepare records in three clear stages, and we label them honestly:

1. First draft (assisted synthesis)

   First-pass entries are seeded from public scholarly sources, gazetteers, government and ASI records, temple-trust publications, ethnographic work, and tradition-specific texts. Language models help structure these materials into drafts. A draft is a starting point for study — not a finished teaching.

2. Editorial review (source-checked)

   Editors check citations, dates, names, links, images, and obvious errors. This is quality control, not religious authority: an editorially reviewed record is safer to read, but it is still not a guardian-approved teaching.

3. Named guardian review (tradition-specific)

   Only this tier counts as guardian review: a credentialed guardian from the relevant tradition — ordained clergy, religious-studies academic, recognized scholar, or community elder — has read, edited, and signed off. Their name, title, and date appear on the entry and on the Advisory Council page.

Today, 1117 of 1,666 records have completed editorial review, and 0 of those carry a named guardian signature; the remaining 549 are provisional drafts under continuous care. The gap is real, the labelling is honest, and our priority is deepening trust rather than only expanding the corpus.

Every entry in the encyclopedia carries one of three review states. We label them honestly so you always know what kind of record you are reading:

0 human-guardian signed. Every reviewed entry to date carries editorial review by the ELGODS Editorial Council — not a named tradition guardian. We are actively recruiting credentialed reviewers. See the Advisory Council page or the review queue to help close this gap.

Every number below is computed from the corpus itself by scripts/citation-coverage.mjs and refreshed on each measurement (2026-10-13). Nothing here is rounded in our favour: an entry counts as cited only if its own frontmatter names at least one Tier-1 or Tier-2 source. Entries citing only a named work with no tier, and entries citing nothing, are counted separately rather than folded into the favourable figure.

That is a claim about provenance — the entry names a source that can be looked up. It is a different question from the one a reader actually asks: which source supports the sentence I am reading? We measure that separately, and where it is weak we say so.

The gap between the two measures

What we corrected, and what it was worth

Our source lists used to carry slots that named no checkable work — the commonest being Multiple District Gazetteer (state-specific), which names no district, no volume, no author and no year, and was written tier: 2, so it was counted here as Tier-2 provenance. Every one of them has now been removed and the gap recorded on the page rather than papered over. 0 remain.

We want to be exact about what that was worth, because it is easy to claim more than it deserves. 0 entries ever rested on one for their Tier-1/2 claim — every placeholder sat alongside real sources, so removing all of them left the headline figure at 84.7% rather than anywhere near the 84.7% it would have dropped to if they had been carrying it. The correction cost us nothing in the measurement and quite a lot in honesty about what our sourcing actually was.

The same audit corrected a measurement error that cut the other way, and this one did move a number. We had been publishing that 0 entries carry no sources at all. That figure was wrong: those entries named Tier-1 and Tier-2 works in their source lists, and our own reading script could not parse them, because it converted apostrophes inside titles into string delimiters and then failed to match anything. The entries were never unsourced. The bucket is genuinely empty now, and the Tier-1/2 figure is 84.7% as a result — higher than we had published, which is the honest direction but still a correction.

We also kept three citations that look like the ones we removed. A gazetteer named Chamoli / Garhwal District Gazetteer with its volume and year unconfirmed names a real work; only its metadata is thin. Deleting it would have discarded real provenance while appearing to fix something, so those are disclosed as incomplete rather than deleted.

84.7% of live entries name a Tier-1 or Tier-2 source. 79 of 1,667 — 4.7% — carry an inline [S#] marker in the body, so a reader can click from a specific paragraph to the specific source behind it. That second figure is small, and we are not going to blur it.

Most of that citation work is in stories, where 35 of 181 entries cite inline. The deity and temple pages are the opposite case: well sourced in their frontmatter, and with almost no source a reader can follow from a sentence. We treat a missing inline citation as a gap to be closed, never as a source to fabricate — a missing citation costs us precision, an invented one costs us the encyclopedia.

By category

The spread is uneven, and we show it rather than hide it. Stories, timeline, travellers, glossary and divine-animals records are fully cited. Temples and practices are the weakest and carry the most work.

Two different gaps, two different fixes

The 236 entries that cite a named work but never declare its tier are not unsourced — the citation is there and the work is usually a well-known scripture. What is missing is the provenance grade. That is mechanical work and we are doing it.

The 0 entries with no sources at all are a different problem. Most are temple records where nothing has been verified yet. Those need research, and they need a person who knows the shrine — which is why they sit on the review queue rather than in an automated pass.

A citation is a claim of provenance, not a certificate of accuracy. Tier-1 and Tier-2 sources are locatable and checkable; what still needs a human is a tradition-holding guardian confirming that the cited work supports the specific claim made. That second step is what the Advisory Council exists to do, and it is tracked separately on the review queue.

These are the largest groups of reviewed-but-unsigned records. If you belong to one of these traditions — or can introduce us to a guardian who does — this is where help matters most.

• Hindu

  660 reviewed records awaiting a named guardian.

• Hinduism

  130 reviewed records awaiting a named guardian.

• Buddhist

  47 reviewed records awaiting a named guardian.

• Christian

  34 reviewed records awaiting a named guardian.

• stories

  28 reviewed records awaiting a named guardian.

• Islamic

  26 reviewed records awaiting a named guardian.

• sacred-texts

  25 reviewed records awaiting a named guardian.

• Jain

  20 reviewed records awaiting a named guardian.

Offer review through the contribution form or the Advisory Council page.

The sanctuary began with the Dharmic traditions of Bhārata (Hindu, Buddhist, Jain, Sikh) and their living folk and tribal layers, and is expanding — slowly and conservatively — into the wider world: Zoroastrian yazatas, Shinto kami, Taoist immortals, the Manifestations of the Bahá'í Faith, and more.

Four rules govern how sacred matter is handled here — in the encyclopedia and in the game alike:

1. Emic names first

   Kathā, sthalapurāṇa, itihāsa, charitra, janamsakhi — each telling keeps its own name. “Myth” and “legend” are used only in comparative contexts, with explanation, never as a quiet demotion of living faith.

2. No caste or gender ranking

   No frame ranks communities by birth; caste-exclusionary violence is named where sources support it, and Dalit, Bahujan, women’s, and folk tellings are preserved as tellings — not footnotes.

3. Contested places, multiple stakeholders

   Where shrines, mosques, churches, monasteries, or sacred groves carry competing claims, every stakeholder community is represented and triumphalism is refused.

4. Pastoral care in play

   The Ramayana game follows Valmiki’s telling while marking its hardest episodes — Tataka, Shurpanakha, Vali, the fire ordeal — as lessons to sit with, not violence to celebrate. Compassion-first choices are always named as dharma.

The full source and content live at github.com/pmaero-byte/ELGODS. Contributions, corrections, and tradition-specific reviewers are welcome — the easiest path is the web contribution form.

We treat every tradition with equal respect. ELGODS does not promote, rank, or syncretize traditions — it documents them as practitioners and scholars within each tradition describe them.