Trifle CLI | Trifle

Everything you need in a single binary

Trifle CLI is a Swiss Army knife for your metrics. Query, push, and analyze data without leaving the terminal.

Query Metrics

Pull time-series data for any key. Filter by granularity, date range, and value path.

Push Data

Track events directly from the terminal or shell scripts. Pipe-friendly for automation.

Aggregate

View aggregated totals across time ranges. Perfect for quick analysis.

Timeline View

See your metrics over time in a clean table format right in your terminal.

Category View

Break down hierarchical data by categories. See country, channel, or any nested dimension.

Multiple Formats

Output as table (human-friendly), JSON (for APIs), or CSV (for spreadsheets). Pipe into jq, awk, or any tool.

Key Feature

AI agents that understand your metrics

Run trifle mcp to start an MCP server. AI coding agents like Claude, Cursor, or Copilot can query your metrics while having full context of your code.

Context-aware analysis
Agents see your code and your metrics at the same time
Natural language queries
Ask "how did orders trend last week?" and get real answers
Works with Claude, Cursor, Copilot
Standard MCP protocol works with any compatible agent
Agent Skills included
Open-source skills teach agents CLI best practices like push patterns, query strategies, and local analytics workflows

{
  "mcpServers": {
    "trifle": {
      "command": "trifle",
      "args": ["mcp"]
    }
  }
}

Add this to your Claude, Cursor, or Copilot MCP configuration and your AI agent can query Trifle metrics directly.

$ trifle track --key tests \
    --values '{"passed": 42, "failed": 3}'

Tracked successfully.
Driver: sqlite (./trifle.db)
Key:    tests
At:     2026-02-14T10:30:00Z

$ trifle metrics timeline --key tests \
    --format table

at                                passed  failed
--------------------------------  ------  ------
2026-02-14T10:00:00.000000+00:00  42      3

Local SQLite driver, no server needed

Run any Trifle command and it automatically creates a local SQLite database. No configuration files, no server setup, no environment variables. It just works.

Automatic database creation

A trifle.db file is created in your working directory on first use

Perfect for local development

Track metrics during development without setting up infrastructure

AI-friendly analytics

Agents use the local SQLite driver to aggregate large datasets that don't fit in context, then query the results

Your data, your format

Same data, three output formats. Pick what fits your workflow or pipe it into other tools.

Table --format table

at                  count
------------------  -----
2026-02-09T07:00Z   246
2026-02-09T08:00Z   112
2026-02-09T09:00Z   88
2026-02-09T10:00Z   53

JSON --format json

[
  {
    "at": "2026-02-09T07:00Z",
    "count": 246
  },
  {
    "at": "2026-02-09T08:00Z",
    "count": 112
  }
]

CSV --format csv

at,count
2026-02-09T07:00Z,246
2026-02-09T08:00Z,112
2026-02-09T09:00Z,88
2026-02-09T10:00Z,53

Pipe JSON into jq, CSV into awk, or table straight to your eyes.

Install in seconds

Choose your preferred installation method. Single binary, no dependencies.

# Install via Homebrew
$ brew install trifle-io/trifle/trifle

# Install with Go
$ go install github.com/trifle-io/trifle-cli@latest

# Download binary
$ curl -sSL https://get.trifle.io | sh

Three ways to connect

Through Trifle App, directly to your database, or fully offline with local SQLite. Pick what fits your setup.

Trifle App

Cloud or self-hosted. Configure your Project or Database in Trifle App, then query them from CLI without exposing direct database access.

Manage dashboards, alerts and monitors.

$ trifle config set driver api
$ trifle config set api-token $TOKEN

Database

Talk to your database directly. No middleman, no proxy. If you have access to the host, point CLI at it and go.

PostgreSQL, MySQL, MongoDB, and Redis.

$ trifle metrics get --driver postgres \
  --dsn $DATABASE_URL \
  --key orders

AI Friendly

Local

Zero-config SQLite driver. No account, no server, no internet required. Great for development, prototyping, and AI agents analyzing large datasets locally.

Everything stays on your machine.

# No config needed. Just run:
$ trifle track --key deploys --values '{"count":1}'

Install in 3.6 seconds

One command to install. One command to track. Zero config required.

$ brew install trifle-io/trifle/trifle

Read the Docs View on GitHub

Single binary | Zero dependencies | Elastic License v2

Your metrics, one command away.