Query DuckDB Examples

Worked examples of querying minitrace archives with the DuckDB backend

Sections

Terminology & Glossary
📖 Documentation
Navigation
20 sectionsv0.1
📄 Query DuckDB Examples — glaze help query-duckdb
query-duckdb

Query DuckDB Examples

Worked examples of querying minitrace archives with the DuckDB backend

Exampleminitraceduckdbglazedquery duckdb

This page shows concrete query examples against converted minitrace archives. For the full flag reference, see go-minitrace help query-commands. For guidance on writing your own queries, see go-minitrace help writing-duckdb-queries.

All examples assume an archive at ./output/active/*/*.minitrace.json. Adjust the --archive-glob for your setup.

One annotation-specific nuance is worth stating up front: go-minitrace query duckdb reads the .minitrace.json archive files it loads. If you created or edited annotations through go-minitrace annotate ..., run go-minitrace annotate sync --output-dir ... first so those annotation changes are present in the archive.

One DuckDB-specific nuance is also worth stating before the examples: JSON arrow operators (-> / ->>) have low precedence. Inside predicates such as WHERE, AND, and OR, parenthesize JSON-arrow extractions so the expression is grouped the way you intend.

Using built-in presets

List all sessions sorted by start time:

go-minitrace query duckdb \
  --archive-glob './output/active/*/*.minitrace.json' \
  --preset session-list

Compare frameworks side by side:

go-minitrace query duckdb \
  --archive-glob './output/active/*/*.minitrace.json' \
  --preset framework-summary \
  --output json

See how tools are used across frameworks:

go-minitrace query duckdb \
  --archive-glob './output/active/*/*.minitrace.json' \
  --preset tool-operation-breakdown

Custom SQL examples

Count sessions by model

go-minitrace query duckdb \
  --archive-glob './output/active/*/*.minitrace.json' \
  --sql "
    SELECT
      environment->>'model' AS model,
      COUNT(*) AS sessions,
      ROUND(AVG(CAST(metrics->>'tool_call_count' AS INT)), 1) AS avg_tools
    FROM sessions_base
    GROUP BY model
    ORDER BY sessions DESC
  "

Token usage by framework

go-minitrace query duckdb \
  --archive-glob './output/active/*/*.minitrace.json' \
  --sql "
    SELECT
      environment->>'agent_framework' AS framework,
      ROUND(SUM(CAST(metrics->>'total_input_tokens' AS BIGINT)) / 1e6, 1) AS input_M,
      ROUND(SUM(CAST(metrics->>'total_output_tokens' AS BIGINT)) / 1e6, 1) AS output_M,
      ROUND(SUM(CAST(metrics->>'total_cache_read_tokens' AS BIGINT)) / 1e6, 1) AS cache_M
    FROM sessions_base
    GROUP BY framework
  "

Most active tools

go-minitrace query duckdb \
  --archive-glob './output/active/*/*.minitrace.json' \
  --sql "
    SELECT
      REPLACE(CAST(json_extract(tc, '\$.tool_name') AS VARCHAR), '\"', '') AS tool,
      COUNT(*) AS invocations
    FROM sessions_base, UNNEST(tool_calls) AS t(tc)
    GROUP BY tool
    ORDER BY invocations DESC
    LIMIT 20
  "

Sessions by hour of day

go-minitrace query duckdb \
  --archive-glob './output/active/*/*.minitrace.json' \
  --sql "
    SELECT
      CAST(timing->>'hour_of_day' AS INT) AS hour,
      COUNT(*) AS sessions
    FROM sessions_base
    WHERE (timing->>'hour_of_day') IS NOT NULL
    GROUP BY hour
    ORDER BY hour
  "

Find long-running sessions

go-minitrace query duckdb \
  --archive-glob './output/active/*/*.minitrace.json' \
  --sql "
    SELECT
      id, title,
      ROUND(CAST(timing->>'duration_seconds' AS DOUBLE) / 3600, 1) AS hours,
      CAST(metrics->>'turn_count' AS INT) AS turns,
      CAST(metrics->>'tool_call_count' AS INT) AS tools
    FROM sessions_base
    WHERE (provenance->>'source_format') NOT LIKE '%subagent%'
    ORDER BY hours DESC
    LIMIT 10
  "

Filter to non-subagent sessions only

go-minitrace query duckdb \
  --archive-glob './output/active/*/*.minitrace.json' \
  --sql "
    SELECT COUNT(*) AS main_sessions
    FROM sessions_base
    WHERE (provenance->>'source_format') NOT LIKE '%subagent%'
  "

Using a SQL file

Save a query to a file:

-- my-query.sql
SELECT
  environment->>'agent_framework' AS framework,
  environment->>'model' AS model,
  COUNT(*) AS sessions
FROM sessions_base
GROUP BY framework, model
ORDER BY sessions DESC;

Run it:

go-minitrace query duckdb \
  --archive-glob './output/active/*/*.minitrace.json' \
  --sql-file ./my-query.sql

Combining with external tools

Pipe JSON output to jq for further filtering:

go-minitrace query duckdb \
  --archive-glob './output/active/*/*.minitrace.json' \
  --preset session-list --output json \
  | jq '[.[] | select(.turns > 50)]'

Export to CSV for spreadsheets:

go-minitrace query duckdb \
  --archive-glob './output/active/*/*.minitrace.json' \
  --preset framework-summary --output csv > summary.csv

See also

  • go-minitrace help js-api-reference — use JavaScript command handlers with mt.db() and normalized SQLite tables when you need reusable multi-query logic
  • go-minitrace help analysis-guide — end-to-end workflow with preset guidance and output format examples
  • go-minitrace help structured-query-commands — promote useful queries into named, reusable structured commands with typed flags
  • go-minitrace help duckdb-query-recipes — more ready-to-use query examples
  • go-minitrace help writing-duckdb-queries — DuckDB JSON syntax, UNNEST, casting, and annotation joins
  • go-minitrace help query-commands — full flag reference
  • go-minitrace help annotation-playbook — correct sync-first workflow when querying annotations created through the CLI
  • go-minitrace help getting-started — step-by-step tutorial that uses query duckdb from the start