Sequence
Understand what happened first, next and last across tools.
local-first work context
Turn work into
AI context
OpenWorkGraph observes app and browser activity locally, preserving the sequence and effort behind real work. Export a session and give that context to ChatGPT, Claude, Gemini, Grok or another AI tool to understand the workflow and spot opportunities to improve or automate it.
$
curl -fsSL https://raw.githubusercontent.com/KAVentures/openworkgraph/main/install.sh | bashdirect GitHub Release ZIP · runs locally · no account required · early public prototype
what it does
See the process behind the output.
Documents show what was saved. OpenWorkGraph captures what happened around them, then lets you export that session as context for the AI you already use.
Effort
See where time, clicks, scrolling and aggregate typing effort accumulate.
Ownership
Keep the evidence local and export it when you choose.
project status
Useful now. Intentionally unfinished.
Desktop observer
Apps, windows, timing, clicks, scrolling and aggregate keypress effort.
AvailableBrowser context
Tab navigation and semantic browser activity.
AvailableLocal export
Export a session to ChatGPT, Claude, Gemini, Grok or your preferred AI tool for analysis.
AvailableMCP / agent access
Direct context access for AI assistants.
In progressTeam deployment
Managed setup and organization-wide context.
In progressAI recommendations
Suggestions for repeated work and automation opportunities.
In progressquestions
Small print, in plain English.
Does it record what I type?+
No. It can count aggregate keypress activity, but it does not record key identities, key order, typed field values or clipboard contents.
Where does the data go?+
The public prototype is local-first. You decide what to export.
Is this employee scoring software?+
No. It is designed to reconstruct work processes and context, not rank people.
Can I use it with AI now?+
Yes. Today you can export a session and give it to ChatGPT, Claude, Gemini, Grok or another AI tool. Direct MCP / agent access is still in progress.
start with the observer
See your work
as context.
Run a real session, export it, and ask your AI where time goes, what repeats, and what might be worth improving or automating.