Ross ROSS = Recommend OSS · open-source software intelligence for agents

OpenAdaptAI/OpenAdapt

For EHRs, payer portals, and any app your team clicks through. A person or an agent shows the task; OpenAdapt turns it into a local program that repeats the steps with no model calls, checks each save, and stops when something doesn't match. observed · 2026-10-21

github.com/OpenAdaptAI/OpenAdapt · homepage · Python · MIT (permissive) observed · 2026-10-21

Health v2 · maintenance only

94/100

  • Activity 100
  • Release rhythm 88
  • Longevity 91
How is this computed?

round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10) — computed 2026-10-21. Adoption (stars, forks) is never an input.

  • gap_med: 0
  • age_days: 1287
  • days_rel: 0
  • days_push: 0
  • n_releases_24m: 64

Full methodology

Adoption not part of the score

1784 stars · 266 forks observed · 2026-10-21

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

OpenAdapt is a Python framework that compiles a demonstrated GUI workflow into an inspectable, deterministic, locally executable program for browser, desktop, RDP, or Citrix/VDI surfaces. Healthy runs make zero generative-model calls, verify declared results before reporting success, and halt with evidence instead of guessing when verification fails.

Use cases

  • automate repetitive data entry into legacy desktop apps with no API
  • record a GUI task once and replay it deterministically
  • automate workflows over RDP or Citrix virtual desktops
  • verify that automated browser actions actually produced the expected result
  • automate patient data entry in healthcare systems with audit evidence
  • replace manual RPA scripting by demonstrating the task instead of coding it

When to choose

  • the task is repeated GUI work on an application with no practical write API
  • you need proof that each consequential step succeeded before reporting success
  • you want local-first, MIT-licensed automation with no model calls on healthy runs
  • a wrong action has operational, financial, or compliance cost and halting is preferable to guessing

When to avoid

  • you need fully autonomous agents that improvise novel UI interactions
  • the workflow has no stable, demonstrable structure or changes constantly
  • you only need simple API-to-API integration without any GUI surface
  • you require a production-admitted, qualified deployment today (admission state is currently not actively admitted)

Facets

framework · maturity active

workflow-automation agent-framework computer-vision gui testing developer-tools healthcare self-hosted python cross-platform windows cli self-hosted rpa gui-automation desktop-automation browser-automation computer-use-agents deterministic-replay automation-from-demonstration rdp citrix process-automation verification local-first automation macos linux

6 sources

Member repositories

RepositoryRoleHealth v2
OpenAdaptAI/OpenAdaptmain94

For agents

markdown · JSON · MCP: product_card(name="OpenAdaptAI/OpenAdapt")

Data as of 2026-10-19T08:05:51.212177+00:00 · Report a problem