WakaTime

Workbench integrates with WakaTime to track your coding activity. Agent sessions are reported as AI coding and terminal commands as human coding, so your WakaTime dashboard shows how much time you spend with AI assistance versus manual work.

Setup

1. Sign up at wakatime.com and copy your API key from Settings.

2. Recommended: Create ~/.wakatime.cfg with your API key — this is shared by all WakaTime plugins:

   [settings]
   api_key = waka_your-api-key-here

3. Open Settings → WakaTime in Workbench and enable the toggle. If you used ~/.wakatime.cfg, the key is detected automatically.

   If you prefer not to create the config file, paste your API key directly into the Settings field.

What gets tracked

Activity	WakaTime category	Example
Agent chat turns	AI coding	Claude editing files, running tools
Terminal commands	Coding	cargo test, git commit, shell commands

Metadata

Workbench sends the following metadata with each heartbeat:

Field	Source	Notes
Project	Repository name from Settings	Falls back to the directory name if the repository can’t be resolved
Language	Detected from tracked files via git ls-files	Extension-based detection, cached for 10 minutes per repository
Machine	System hostname	Omitted if hostname is unavailable
OS	Build-time platform	Mac, Linux, or Windows
Prompt count	Cumulative turns per chat session	AI coding heartbeats only
Editor	Always “Workbench”	Sent via the User-Agent header

Custom API URL

For self-hosted WakaTime servers, expand the Advanced section in Settings and set your API endpoint.

Privacy

• Only the repo directory path, project name, language name, and branch are sent — no file contents, filenames, command text, or terminal output.
• Hostname is sent as a machine identifier, consistent with other WakaTime plugins.
• API keys are never logged, included in error events, or sent to the frontend.