Querying Events#

Vibetracker stores events in SQLite at ~/.vibetracker/events.db. You can query this data using the CLI or directly with SQLite tools.

Using the CLI#

The query command executes SQL and returns JSON:

vibetracker query "SELECT * FROM events LIMIT 5"

Direct Database Access#

You can also use any SQLite client:

sqlite3 ~/.vibetracker/events.db "SELECT * FROM events LIMIT 5"

Or with Bun:

import { Database } from "bun:sqlite"

const db = new Database(process.env.HOME + "/.vibetracker/events.db")
const events = db.query("SELECT * FROM events LIMIT 5").all()

Common Queries#

Session Statistics#

Total sessions and events:

SELECT
  COUNT(DISTINCT session_id) as sessions,
  COUNT(*) as total_events
FROM events

Sessions per day:

SELECT
  DATE(timestamp) as day,
  COUNT(DISTINCT session_id) as sessions
FROM events
GROUP BY DATE(timestamp)
ORDER BY day DESC

Average session duration:

SELECT
  AVG(session_duration_ms) / 1000.0 / 60.0 as avg_minutes
FROM events
WHERE event_type = 'session_end'
  AND session_duration_ms IS NOT NULL

Token Usage#

Total tokens by model:

SELECT
  model,
  SUM(prompt_tokens) as prompt_tokens,
  SUM(completion_tokens) as completion_tokens,
  SUM(total_tokens) as total_tokens
FROM events
WHERE total_tokens IS NOT NULL
GROUP BY model
ORDER BY total_tokens DESC

Tokens per session:

SELECT
  session_id,
  SUM(total_tokens) as tokens,
  COUNT(*) as turns
FROM events
WHERE event_type = 'turn_end'
GROUP BY session_id
ORDER BY tokens DESC
LIMIT 10

Daily token usage:

SELECT
  DATE(timestamp) as day,
  SUM(total_tokens) as tokens
FROM events
WHERE total_tokens IS NOT NULL
GROUP BY DATE(timestamp)
ORDER BY day DESC

Tool Usage#

Tool call frequency:

SELECT
  tool_name,
  COUNT(*) as calls
FROM events
WHERE event_type = 'tool_call'
GROUP BY tool_name
ORDER BY calls DESC

Tool success rate:

SELECT
  tool_name,
  COUNT(*) as total,
  SUM(CASE WHEN tool_success = 1 THEN 1 ELSE 0 END) as succeeded,
  ROUND(100.0 * SUM(CASE WHEN tool_success = 1 THEN 1 ELSE 0 END) / COUNT(*), 1) as success_rate
FROM events
WHERE event_type = 'tool_call'
GROUP BY tool_name
ORDER BY total DESC

Average tool duration:

SELECT
  tool_name,
  ROUND(AVG(tool_duration_ms), 0) as avg_ms,
  COUNT(*) as calls
FROM events
WHERE event_type = 'tool_call'
  AND tool_duration_ms IS NOT NULL
GROUP BY tool_name
ORDER BY avg_ms DESC

Slowest tool calls:

SELECT
  tool_name,
  tool_input,
  tool_duration_ms
FROM events
WHERE event_type = 'tool_call'
  AND tool_duration_ms IS NOT NULL
ORDER BY tool_duration_ms DESC
LIMIT 10

File Operations#

Most edited files:

SELECT
  file_path,
  COUNT(*) as edits,
  SUM(file_lines_added) as lines_added,
  SUM(file_lines_removed) as lines_removed
FROM events
WHERE file_path IS NOT NULL
GROUP BY file_path
ORDER BY edits DESC
LIMIT 20

File operations by type:

SELECT
  file_action,
  COUNT(*) as count
FROM events
WHERE file_action IS NOT NULL
GROUP BY file_action

Net lines of code changed:

SELECT
  SUM(COALESCE(file_lines_added, 0)) as added,
  SUM(COALESCE(file_lines_removed, 0)) as removed,
  SUM(COALESCE(file_lines_added, 0)) - SUM(COALESCE(file_lines_removed, 0)) as net
FROM events
WHERE file_path IS NOT NULL

Git Context#

Sessions by repository:

SELECT
  session_git_repo,
  COUNT(DISTINCT session_id) as sessions,
  COUNT(*) as events
FROM events
WHERE session_git_repo IS NOT NULL
GROUP BY session_git_repo
ORDER BY sessions DESC

Activity by branch:

SELECT
  session_git_branch,
  COUNT(DISTINCT session_id) as sessions
FROM events
WHERE session_git_branch IS NOT NULL
GROUP BY session_git_branch
ORDER BY sessions DESC
LIMIT 20

Error Analysis#

Error frequency:

SELECT
  error_code,
  error_message,
  COUNT(*) as occurrences
FROM events
WHERE event_type = 'error'
GROUP BY error_code, error_message
ORDER BY occurrences DESC
LIMIT 20

Sessions with errors:

SELECT
  session_id,
  COUNT(*) as error_count
FROM events
WHERE event_type = 'error'
GROUP BY session_id
ORDER BY error_count DESC
LIMIT 10

Agent Source Comparison#

Events by source:

SELECT
  source,
  COUNT(*) as events,
  COUNT(DISTINCT session_id) as sessions
FROM events
GROUP BY source

Token usage by source:

SELECT
  source,
  SUM(total_tokens) as tokens,
  COUNT(DISTINCT session_id) as sessions,
  ROUND(1.0 * SUM(total_tokens) / COUNT(DISTINCT session_id), 0) as tokens_per_session
FROM events
WHERE total_tokens IS NOT NULL
GROUP BY source

Subagent Analysis#

Subagent usage by type:

SELECT
  agent_type,
  COUNT(*) as spawned
FROM events
WHERE agent_type IS NOT NULL
GROUP BY agent_type
ORDER BY spawned DESC

Filtering Tips#

By Time Range#

-- Last 24 hours
WHERE timestamp > datetime('now', '-1 day')

-- Last 7 days
WHERE timestamp > datetime('now', '-7 days')

-- Specific date range
WHERE timestamp BETWEEN '2025-01-01' AND '2025-01-31'

-- Today only
WHERE DATE(timestamp) = DATE('now')

By Session#

-- Specific session
WHERE session_id = 'abc123'

-- Sessions in a repository
WHERE session_git_repo LIKE '%vibetracker%'

By Event Type#

-- Only tool calls
WHERE event_type = 'tool_call'

-- Session boundaries
WHERE event_type IN ('session_start', 'session_end')

Table Schema#

The events table contains all fields from the Event Schema. Columns map directly to event fields with the same names.

Indexes available for performance:

Index	Columns
`idx_events_timestamp`	`timestamp`
`idx_events_session`	`session_id`
`idx_events_synced`	`synced_at`
`idx_events_agent`	`agent_id`