메뉴
Use when combining vector (semantic) and full-text (lexical) search results into one ranking in a Go app with gosqlite — Reciprocal Rank Fusion via the fusion package.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when the task needs a SQL function or virtual table from the gosqlite ext/ catalog — regexp, uuid, hash, ipaddr, zorder, stats, unicode, encode, text, fuzzy, decimal, money, time, eval, csv, lines, statement, closure, pivot, rtree, series, array, bloom, spellfix1, fileio, blobio.
Use when debugging surprising behaviour with gosqlite, or as a pre-ship checklist — the gotchas around per-conn hooks, the connection pool, vec quirks, bind limits, and what is not supported.
Use when the task needs encryption at rest, storing a database compressed at rest, page-checksum corruption detection, an in-memory database with isolation, or serving a database from an embed.FS / byte buffer in a Go app with gosqlite.
Use when using gorm.io/gorm with gosqlite — the gosqlite.org/gorm dialector, a CGo-free drop-in for glebarez/sqlite and gorm.io/driver/sqlite, shipped as its own module.
Use when switching an existing Go project to gosqlite from mattn/go-sqlite3, modernc.org/sqlite, glebarez/sqlite, gorm.io/driver/sqlite, or ncruces/go-sqlite3.
Use when opening a SQLite database in Go with gosqlite.org, choosing the "sqlite" vs "sqlite3" driver name, or deciding which sub-package to reach for. The starting point for any task using this driver.
| name | hybrid-search |
| description | Use when combining vector (semantic) and full-text (lexical) search results into one ranking in a Go app with gosqlite — Reciprocal Rank Fusion via the fusion package. |
The fusion package merges two ranked result sets via Reciprocal Rank Fusion (Cormack 2009). Pure Go, no SQLite dependency.
import "gosqlite.org/fusion"
vecHits, _ := tbl.KNNSlice(ctx, queryVec, 50) // see vector-search
ftsHits, _ := idx.SearchSlice(ctx, fts.Term("brown fox"), fts.WithLimit(50)) // see full-text-search
vecKeys := make([]int64, len(vecHits))
for i, h := range vecHits { vecKeys[i] = h.Rowid }
ftsKeys := make([]int64, len(ftsHits))
for i, h := range ftsHits { ftsKeys[i] = h.Key }
top, _ := fusion.RRF([][]int64{vecKeys, ftsKeys}, fusion.WithLimit(20))
for _, r := range top { fmt.Println(r.Key, r.Score) }
fusion.RRF([][]K{...}, opts...) — fuse any number of ranked key lists.fusion.RRF2(a, b) — convenience for exactly two lists.int64 rowid or string).Run the two searches with a generous limit (50–100) each, fuse, then take the top N. Prerequisites: the vector-search and full-text-search skills. Full reference: docs/guides/hybrid-search.md.