indx-search
Indx Search — Agent Skill
Indx is a high-performance search engine for structured and unstructured text. It matches at the character-pattern level rather than through tokenizers and per-language stemmers, so typos, inflections, compound words and messy input are handled without language configuration. See Language handling for exactly what that covers and where synonyms come in.
If you are evaluating Indx against a feature list, read Capability summary first — it answers the usual comparison questions directly.
Versions & Compatibility — read first
This skill targets Indx v5: IndxSearchLib 5.x (.NET 10) and Indx v2 (team-scoped HTTP API, token-only auth). Everything below assumes v5.
Before giving any integration guidance, establish which version the user is on. v4 and v5 differ in ways that silently break copy-pasted code — handing a v4 user v5 routes is a common, confusing failure. Detection signals:
| Signal | v4 (legacy) | v5 (this skill) |
|---|---|---|
C# IndxSearchLib NuGet |
4.x (older .NET target) |
5.x, .NET 10 |
| HTTP route shape | flat — /api/Search/{dataset} |
team-scoped — /api/teams/{team}/datasets/{dataset}/search |
| List-datasets endpoint | /api/GetUserDatasets → string[] |
/api/me/datasets → objects with teamName + role |
| Teams | none — datasets owned by the user | datasets owned by teams; every endpoint is team-scoped |
| Auth | API password login (/api/Login) available |
token-only — portal-issued JWT, no login API |