Does Chroma support keyword or hybrid search?

No. Chroma supports vector (ANN) search and grep (substring and regex patterns via where_document). BM25 keyword ranking and hybrid search are not available. Chroma is primarily a developer tool for vector search workflows.

Can I use regex patterns with Chroma grep?

Yes. Chroma passes the pattern to where_document.$regex directly. Regex evaluation can be slow on large collections — prefer vector or substring search for performance-sensitive paths.

What metadata fields does Chroma expose for policies?

Chroma stores arbitrary key-value metadata per document. Gateco reads any field (classification, owner_id, etc.) via inline metadata resolution without maintaining a separate sidecar store.

Vector DBTier 2

Chroma

Add RBAC and audit logging to Chroma embedding database collections.

Search capabilities

Vector (ANN)	Keyword (ranked)	Keyword (filter)	Hybrid (native)	Grep (substring)	Grep (regex)

Chroma in-process is fastest; remote HTTP mode adds network overhead. Policy overhead <25ms p95.

sidecar metadatainline metadata

Chroma is a developer-friendly embedding database popular for local development and rapid prototyping. Gateco connects to Chroma as a Tier 2 retrieval connector, adding policy enforcement and audit logging to Chroma collections.

Chroma's where_document filter supports $contains (substring) and $regex (regular expression) patterns. Both are available as Gateco grep modes. Chroma does not have ranked keyword search or hybrid search.

Chroma is one of the connectors that supports regex grep natively: patterns are passed to where_document.$regex directly. Be aware that regex evaluation in Chroma can be slow on large collections.

Chroma stores arbitrary key-value metadata alongside each document. Gateco uses inline metadata resolution to read classification, owner_id, or any other field from the Chroma document metadata — no separate sidecar store required.

Sample search config

{
  "collection_name": "documents"
}

Sample policy

{
  "name": "Developer access",
  "effect": "allow",
  "rules": [{
    "conditions": [
      {"field": "principal.groups", "operator": "contains", "value": "developers"},
      {"field": "resource.classification", "operator": "in", "value": ["internal", "public"]}
    ]
  }],
  "selectors": [{"connector_type": "chroma"}]
}

Policy conditions reference resource.* and principal.* fields. Policy reference →

Frequently asked questions

Does Chroma support keyword or hybrid search?: No. Chroma supports vector (ANN) search and grep (substring and regex patterns via where_document). BM25 keyword ranking and hybrid search are not available. Chroma is primarily a developer tool for vector search workflows.
Can I use regex patterns with Chroma grep?: Yes. Chroma passes the pattern to where_document.$regex directly. Regex evaluation can be slow on large collections — prefer vector or substring search for performance-sensitive paths.
What metadata fields does Chroma expose for policies?: Chroma stores arbitrary key-value metadata per document. Gateco reads any field (classification, owner_id, etc.) via inline metadata resolution without maintaining a separate sidecar store.

Ready to connect Chroma?

Follow the step-by-step setup guide or talk to the team for help with your specific configuration.

Setup guide →Talk to the team All integrations