Find the right component on the first try.
teleon.dev/docs/search combines keyword ranking, a domain vocabulary, trigram matching and near-duplicate hashing, and picks the right mix for each query. Results come back as capability cards or byte-range links.
Each cell mixes signals differently. Darker cells answered more queries; the router learns from the pattern.
What it does
Eight signals, six modes
Keywords, vocabulary, trigrams, SimHash, MinHash and more, mixed per query.
Short cards for agents
Each hit is a short capability card an agent can read in a few lines.
Misses become work
Queries that find only weak matches are logged as gaps for the probe queue.
The contract
A query returns ranked cards with the cell that answered it, so tuning has evidence.
/v1/docs/search for generated documentshybrid_fts_trigram_indexsimhash_minhash_lsh
Endpoints
| Method | Path | Access | What it does | Status |
|---|---|---|---|---|
GET | /v1/search | RING-3 | Search capability cards | |
GET | /v1/docs/search | RING-3 | Search generated documents | Listed |
GET | /v1/resources/lsh-query | RING-3 | Find near-duplicate resources | Listed |
Listed paths appear in the service’s published files today. In the one-host plan every path here is served under api.teleon.dev/v1/search.
How it connects
For agents
Give your agent the read-only guide first. It reports back before any write.
Words on this page
- BM25F
- A ranking formula for keyword search that weighs matches by field, so a title match counts more than a body match.
- Trigram index
- An index of every three-character sequence, so substring and typo-tolerant search stays fast.
- SimHash
- A short fingerprint of a document where similar documents get similar fingerprints.
- MinHash
- A compact signature for estimating how much two sets overlap, used to find near-duplicates.