--- title: "Introduction" description: "Cadence parses your coding session logs and gives you detailed data around cost, quality, and efficiency." url: "https://teamcadence.ai/docs/" --- ## What Cadence is Cadence helps you continually improve your AI-assisted software engineering. It pulls out important info from your coding session logs and matches it with best practices, so that you can decide if you want to change your process. ## How data gets in 1. You work as normal with your AI coding agent of choice - see [supported agents](/docs/integrations/agents/). 2. The Cadence desktop app's background daemon notices new session activity, resolves it to a git repo, and - if [that repo is enabled for access](/docs/privacy/#what-gets-uploaded) - uploads the session. 3. Before upload the desktop app [sanitises secrets and PII](/docs/security/) from the session. 4. The server analyzes the session (what changed, how much it cost, how the work went) and links it to the commits and pull requests it produced. 5. Results feed the desktop app's local views, and become evidence your local [CLI](/docs/cli/overview/) and [Cue](/docs/cue/overview/) can query. See [Sessions](/docs/concepts/sessions/) for what "a session" actually means, and [Privacy & data](/docs/privacy/) for exactly what is and isn't visible to your org. ## The desktop app and the CLI Cadence has two local surfaces. The [desktop app](/docs/desktop/overview/) is the always-on background piece - it uploads sessions, and its menubar/tray popover gives you quick local context: recent activity, [tune-ups](/docs/desktop/tune-ups/), and [Radar](/docs/desktop/radar/). The [CLI](/docs/cli/overview/) is read-only and built primarily for coding agents to query - your own sessions, reviews, and blame history - as evidence mid-task. ## Where to go next - New to Cadence? Start with [Installation & onboarding](/docs/installation/). - Curious what a session costs and why? Read [Cost & tokens](/docs/concepts/cost-and-tokens/). - Want an AI agent to use your own coding history as evidence mid-task? Read [Cadence Cue](/docs/cue/overview/). - Wiring up Cadence from a script or another agent? Read the [CLI reference](/docs/cli/commands/).
Docs/Introduction
Introduction
Cadence parses your coding session logs and gives you detailed data around cost, quality, and efficiency.
What Cadence is
Cadence helps you continually improve your AI-assisted software engineering. It pulls out important info from your coding session logs and matches it with best practices, so that you can decide if you want to change your process.
How data gets in
- You work as normal with your AI coding agent of choice - see supported agents.
- The Cadence desktop app’s background daemon notices new session activity, resolves it to a git repo, and - if that repo is enabled for access - uploads the session.
- Before upload the desktop app sanitises secrets and PII from the session.
- The server analyzes the session (what changed, how much it cost, how the work went) and links it to the commits and pull requests it produced.
- Results feed the desktop app’s local views, and become evidence your local CLI and Cue can query.
See Sessions for what “a session” actually means, and Privacy & data for exactly what is and isn’t visible to your org.
The desktop app and the CLI
Cadence has two local surfaces. The desktop app is the always-on background piece - it uploads sessions, and its menubar/tray popover gives you quick local context: recent activity, tune-ups, and Radar. The CLI is read-only and built primarily for coding agents to query - your own sessions, reviews, and blame history - as evidence mid-task.
Where to go next
- New to Cadence? Start with Installation & onboarding.
- Curious what a session costs and why? Read Cost & tokens.
- Want an AI agent to use your own coding history as evidence mid-task? Read Cadence Cue.
- Wiring up Cadence from a script or another agent? Read the CLI reference.