Skip to main content
Version: v2 (current)

Features

TL;DR. PanDev Metrics collects engineering signals from four layers (IDE, Git, task tracker, CLI/browser), normalizes them into one timeline per person, and turns that timeline into DORA metrics, real coding hours, and the only cost-of-a-feature calculation on the market. Everything below is available in both Cloud and on-prem — same product.

Data collection​

This is the layer competitors skip. PanDev Metrics emits activity events directly from where engineers actually work.

IDE plugins​

Native plugins across the full IDE landscape:

  • JetBrains family (10 IDEs) — IntelliJ IDEA, PyCharm, WebStorm, PhpStorm, GoLand, Rider, CLion, RustRover, RubyMine, Android Studio.
  • Microsoft — Visual Studio (all versions; 2024+ in theory works, not officially tested), VS Code.
  • VS Code forks — Cursor and Windsurf use the VS Code plugin directly.
  • Xcode — for iOS/macOS teams.
  • Eclipse family — Eclipse and Eclipse-based IDEs.

Plugins update automatically if the IDE supports it, otherwise manually. Offline? The plugin queues events locally and flushes when the engineer is back online.

What the plugin captures: file names, activity events (file changes, scroll, open/close), foreground app. What it never captures: source code, screenshots. AI-tool usage is tracked as two distinct signals — engineer prompting (counts as developer activity) and AI auto-generating (counts as AI activity).

Git platforms​

Connectors for every mainstream hosting:

  • GitHub — both GitHub App (recommended, easier) and OAuth. Minimum permissions: Repository Read, Pull Request Read/Write.
  • GitLab — Cloud and Self-Managed. Minimum scopes: api, write_repository.
  • Bitbucket — Cloud, Server, Data Center.
  • Azure DevOps — Repos, Pipelines, Boards.

Task trackers​

Linking code to business work:

  • Jira — Cloud and Server/Data Center (on-prem). Requires an Administrator account because PanDev creates webhooks on connect.
  • Yandex Tracker — production-ready, OAuth.
  • ClickUp — production-ready.
  • Azure Boards — supported.
  • Coming up: Monday, Trello, Asana.
  • Not supported: Linear.

If you don't want any tracker integration at all, run the platform in no-integrations mode and rely on IDE + Git signals alone.

CLI and browser​

  • CLI plugins for Mac and Linux (Windows in development). Captures only the command name and the fact of execution — a heartbeat of activity. Codex CLI and Claude Code are first-class citizens.
  • Browser extensions for 10 browsers (Chrome, Brave, Edge, Firefox, Opera, Arc, Vivaldi, Yandex, Whale, Sidekick). Org-level whitelist controls what gets tracked; each user keeps a personal blacklist for what to never track.

Engineering analytics​

Once events land in PanDev, they roll up into dashboards.

DORA metrics​

The industry-standard delivery indicators, with performance bands (Elite / High / Medium / Low):

  • Deployment frequency
  • Lead time for changes
  • Change failure rate
  • Mean time to restore

Each metric is drillable down to the underlying events.

Productivity and engagement​

A broader set of indicators tracked at the engineer, team, role and department level. The most distinctive ones:

  • Real coding hours. A gap shorter than 15 minutes between IDE events counts as active time. A longer gap doesn't. This calibration is the foundation of every "hours" number you'll see.
  • Overtime. Any work outside business hours on a weekday. Reflects burnout risk honestly, without manual logging.
  • Delivery Index. Ratio of research time to development time in code — surfaces engineers and teams where the balance has drifted.
  • Focus and deep work. Long uninterrupted IDE blocks, separated from short scattered ones.
  • AI-assisted vs hand-written work. Two distinct signals so AI tooling doesn't inflate or deflate productivity by accident.

Custom metrics are not supported — the indicator set is curated and consistent across tenants.

People and projects​

  • Employee card — profile, activity, work history, projects, time distribution. Everything you need for a Performance Review in one place.
  • Project page — repository as the unit, with contributors, activity, timeline.
  • Org structure — teams nested inside departments, all of it inside a tenant. Roles map to the UI labels: Admin, Maintainer, Viewer at tenant level, plus department-level Owner / Maintainer / Viewer / Finance (Financier).

Cost per feature — our unique edge​

This is the single feature you won't find elsewhere on the market.

PanDev Metrics calculates the direct cost of a task as:

time logged against the task × the engineer's hourly rate + a proportional share of that engineer's unattributed time.

What that buys you:

  • A real dollar number against every Jira/Tracker issue you ship.
  • Aggregated costs by project, team, department, or quarter.
  • A rate change applied retroactively — flip a hourly rate and historical costs recalculate across the entire timeline.

The finance layer supports hourly and monthly salary types, in USD only, with manual public holidays and per-employee working calendars (which override the tenant-level calendar). Reports export to XLSX.

The finance area is gated behind the dedicated Finance (Financier) role — assigned from the admin panel by an Owner.

Cloud vs on-premises​

The product is identical in both modes — same features, same UI, same dashboards. Two differences:

CapabilityCloudOn-prem
Multi-tenant orgsyesno — single tenant per install
Google sign-inyesno (LDAP supported)
Everything elseyesyes

On-premises deployment​

  • Targets: Docker (Compose v2.0+, Engine ≥ 20.10) or Kubernetes (≥ 1.28).
  • Hardware: 8 CPU / 16 GB RAM / 50 GB SSD for app, 16 CPU / 24 GB RAM / 100 GB SSD for the dedicated PostgreSQL host.
  • Database: PostgreSQL 17.
  • Auth: LDAP only for SSO (no SAML/OIDC).
  • Backups: pg_dump is enough.
  • Upgrades: swap image tags in docker-compose.yml or the Helm chart.
  • Ports: 8080 app, 9090 actuator.
  • Self-signed TLS: supported, including the option to disable SSL validation on the client side.
  • Air-gapped installs: not supported.

There is a small amount of outbound traffic on on-prem; it cannot be turned off entirely.

What's coming​

Items the team is actively working on (not described in depth elsewhere yet):

  • Built-in task tracker (Task Flow / Kanban) — our own lightweight tracker so the loop doesn't depend on Jira. In development; the route exists in the navigation but the screen is not yet live in production.
  • Datavisor — a more open data-exploration surface on top of the engineering warehouse.
  • Expanded exports — additional formats and destinations beyond the current XLSX payment report.