tgsonar.com

For AI agents and assistants

This is the machine-readable way into the catalogue: the same data as on the site, without the markup. Take the JSON from the addresses below — there is no need to parse HTML.

11.4K
objects measured
7.7K
in the open index
12.1K
snapshots on record
2026-09-19
last measurement

Endpoints

AddressReturnsDescription
/ai/index.jsonManifestEvery address, the licence and the citation rules
/ai/dataset.jsonDatasetSchema.org Dataset: fields, method, coverage
/ai/entities.jsonObjectsThe whole open index in a single file
/ai/top.jsonTopThe 100 objects with the highest score
/ai/categories.jsonCategoriesVolume, average and maximum score
/ai/facts.jsonFactsSummary figures with the basis stated for each
/ai/entity/<username>.jsonObject cardA full object plus its measurement history
/llms.txtllms.txtA short map of the section

Terms

What you can learn here

Limits of the data

We measure what Telegram shows publicly, and we say plainly what we do not know. The score is a percentile inside a category: using it to weigh a film channel against a trading channel is meaningless. For comparison across categories there are three indexes on an absolute 0–100 scale — trust, usefulness, activity — and confidence: how many independent measurements stand behind the number. One measurement is a hypothesis, not a fact, and the data says so.

What the data does not hold: the content of posts, personal details of authors, judgements about anyone's good faith. An object missing from the catalogue means nothing beyond the fact that we have not reached it yet.

Large categories

Benchmarks by category — the Benchmarks section. The score formula is worked through in the methodology.