Event Schema#

Every event in Vibetracker follows the VibeEvent schema. This page documents all available fields.

Database Overview#

Storage Location#

Vibetracker stores all events in a SQLite database at:

~/.vibetracker/events.db

Table Structure#

All events are stored in a single events table. This is intentionally designed as a wide table similar to observability and analytics databases (like ClickHouse or TimescaleDB schemas). The wide table design allows you to:

• Run complex analytical queries across all event types without joins
• Filter and aggregate on any combination of fields
• Easily export or sync data without reconstructing relationships

Schema Definition#

CREATE TABLE `events` (
	`id` text PRIMARY KEY NOT NULL,
	`timestamp` text NOT NULL,
	`user_id` text NOT NULL,
	`team_id` text,
	`machine_id` text,
	`session_id` text NOT NULL,
	`event_type` text NOT NULL,
	`source` text NOT NULL,
	`session_cwd` text,
	`session_git_repo` text,
	`session_git_branch` text,
	`session_duration_ms` integer,
	`turn_index` integer,
	`prompt_tokens` integer,
	`completion_tokens` integer,
	`total_tokens` integer,
	`model` text,
	`tool_name` text,
	`tool_name_raw` text,
	`tool_input` text,
	`tool_output` text,
	`tool_duration_ms` integer,
	`tool_success` integer,
	`mcp_server` text,
	`mcp_tool_name` text,
	`file_path` text,
	`file_action` text,
	`file_lines_added` integer,
	`file_lines_removed` integer,
	`bash_command` text,
	`bash_command_output` text,
	`error_message` text,
	`error_code` text,
	`prompt_text` text,
	`agent_id` text,
	`agent_type` text,
	`meta` text,
	`synced_at` text
);

CREATE UNIQUE INDEX `idx_events_dedup` ON `events` (`session_id`, `timestamp`, `event_type`, COALESCE(`tool_name_raw`, ''), COALESCE(`tool_input`, ''));
CREATE INDEX `idx_events_timestamp` ON `events` (`timestamp`);
CREATE INDEX `idx_events_session` ON `events` (`session_id`);
CREATE INDEX `idx_events_synced` ON `events` (`synced_at`);
CREATE INDEX `idx_events_agent` ON `events` (`agent_id`);

Field Reference#

Core Fields#

These fields are present on every event.

Optional Core Fields#

Event Types#

The event_type field indicates what kind of event occurred:

Agent Sources#

The source field identifies which agentic coding tool generated the event:

Session Context Fields#

These fields provide context about the coding session.

Turn and Token Fields#

These fields track AI model usage per turn.

Tool Execution Fields#

These fields are populated on tool_call events.

Tool Names#

The tool_name field is normalized across agent sources:

MCP Server Fields#

These fields are populated when an MCP (Model Context Protocol) tool is invoked.

File Operation Fields#

These fields provide details about file operations.

Bash Execution Fields#

These fields capture shell command executions.

Error Fields#

These fields are populated on error events.

Prompt Fields#

Subagent Fields#

These fields track subagent/task spawning.

Metadata Fields#