Search

MemNexus provides powerful search that understands meaning, not just keywords. Search uses a hybrid approach that combines vector similarity with full-text matching for the best results.

Basic search

mx memories search --query "deployment issues last week"

This runs a hybrid search — semantic (vector) and keyword (full-text) — and merges results using Reciprocal Rank Fusion (RRF).

Search modes

Timeline mode

Sort results chronologically with state indicators:

mx memories search --query "API improvements" --timeline

Each result shows an effective state:

• [CURRENT] — Active, valid information
• [SUPERSEDED] — Replaced by newer information
• [CONTRADICTED] — Conflicting information exists

Use timeline mode when reconstructing what happened or understanding how decisions evolved.

Brief mode

Get a compact summary (top 5 results with 80-character previews):

mx memories search --query "database password policy" --brief

Recent results

Limit to a time window:

mx memories search --query "deployments" --recent 24h
mx memories search --query "bug fixes" --recent 7d
mx memories search --query "decisions" --recent 2w

Group by conversation

See results organized by work session instead of flat list:

mx memories search --query "authentication" --group-by conversation

Filtering

Include topics

Return only memories with at least one matching topic:

mx memories search --query "authentication" --topics "implementation,completed"

Exclude topics

Remove memories with any of the specified topics:

mx memories search --query "debugging" --exclude-topics "meeting-notes"

Combine filters

mx memories search \
  --query "authentication" \
  --topics "implementation" \
  --exclude-topics "meeting-notes" \
  --recent 7d \
  --timeline

GraphRAG search

For complex queries that benefit from graph traversal:

mx graphrag query --query "What were the main technical decisions this month?"

Community-based queries

Search across detected topic communities:

mx graphrag query-communities --query "infrastructure changes" --limit 5

Explain a memory

Get graph context for a specific memory:

mx graphrag explain --memory-id <id>

Pagination

mx memories search --query "deployment" --limit 10
mx memories search --query "deployment" --limit 10 --page 1

ID-only mode

Get just the memory IDs for piping:

mx memories search --query "deployment" --id-only

Combine with batch get:

mx memories search --query "deployment" --id-only | mx memories get --stdin

Search tips

• Search by meaning: "that deployment issue" finds memories about deployments even without the exact words
• Search by keyword: "ECONNREFUSED" finds exact matches
• Combine with timeline: See how understanding evolved over time
• Use topic filters: Narrow results when you know the category
• Use recent: Limit to relevant time periods for faster, focused results

Next steps

• Core Concepts — Understand the data model
• How It Works — Learn about the search pipeline