Helix Query Optimization

Optimize Helix queries by aligning their shape with what the Helix interpreter actually does at each step. Every rule in this skill is grounded in the source — see REFERENCE.md for the mechanism and the file:line cite, and EXAMPLES.md for paired Rust DSL + JSON forms.

When To Use

Use this skill when the task is to:

• understand why a Helix query is slow
• review a query for missed index push-down or accidental full scans
• decide if a where_(...) or has(...) is filtering at the index or in memory
• tighten BM25 or vector search routes (tenant scope, distance lifecycle, projection)
• bound Repeat traversals or order Coalesce branches by cost
• audit write-path safety (drop_edge vs drop_edge_by_id, add_n dupes, for_each_param cost)

Reference Files

• REFERENCE.md — full optimizer mental model: index types, source dispatch table, predicate-index resolution catalog, RuntimeState transitions, limit/dedup pushdown, OrderBy paths, projection rules, repeat/branching cost, batch semantics, write-path mechanics, dynamic query cost. Every claim cites enterprise/helix/src/traversal/interpreter.rs (helix-hyperscale) or sdks/rust/src/dsl.rs (the helix-db DSL crate).
• EXAMPLES.md — paired Rust DSL + JSON dynamic patterns for every optimization rule below: weaker form, stronger form, what changed and why.