This is a personal proof-of-concept project. It is not intended for production use. Please refrain from using it.

AI testing & debugging agents

Describe a flow.
It runs it. It debugs it.

Write a user journey in plain English — sign up, log in, check out. The agents run it end to end in your real app, and when a step breaks they trace it to the exact backend line and file the fix. Read-only — self-hosted, your own key.

Source-available · BYO Anthropic key · cloud or self-hosted

What it does

You describe it.
The agents debug it.

Agents that reproduce the bug in a real browser, follow it to the backend line, and file the fix — or probe your infra directly when it isn't the UI. Read-only, always.

01

Agents you debug with

Tell them what's broken in plain English — or fire an autonomous investigation and they root-cause it on their own, across logs, infra, and the UI.

The autonomous agent →
02

Root-caused to the backend line

A failing request's trace_id is stitched to your server logs — the error, the cause, the fix. Past the symptom, to the line that threw.

How root cause works →
03

No OpenTelemetry required

Point it at Loki, Datadog, Elasticsearch, an HTTP endpoint, or a log file. Upliftr stitches the trace itself — no instrumentation project.

Connect logs →
04

They reproduce the failure

The agents drive a real browser and re-ground on the live DOM each run — a confirmed reproduction, not a guess from telemetry.

How it works →
05

Auto-filed, deduped issues

Confirmed failures become root-caused tickets — on the built-in board, or GitHub, GitLab, Jira. Re-runs comment, never duplicate.

Filing issues →
06

Checks that run in CI

Whatever they verify is saved as a plain-English check that re-runs forever — in your IDE (MCP) and as a gate on every PR/MR.

In CI & your IDE →

Agents, not dashboards

Monitoring watches.
Agents investigate.

Sentry and Datadog show you an error after it happened, from telemetry you had to wire up. Upliftr's agents reproduce the failure on demand and follow it across the UI, the API, and the exact backend line. Active, not passive — and read-only.

UI: what the user hit

“Save profile” returned a silent 500. Nothing shown to the user.

Backend: the real cause

NullPointerException in upload.py:142. Avatar saved before the user row commits.

Filed: GitHub #284

Cause + suggested fix, deduped. The ticket was waiting for you.

End-to-end automation

Full user flows,
in plain English.

The agents drive your real app through multi-step journeys — sign up, then log in as that same user, then check out. Cases run in order and share fixtures, and each is kept as readable YAML you can edit. No selectors, no page objects.

Your infra. Your keys. Your data.

It runs where you do.

Every tool like this is closed SaaS — your app and data go to them. Upliftr is source-available and runs entirely in your environment, one command. Bring your own key; nothing leaves your network.

Self-host in one line

curl -fsSL https://get.upliftr.io | bash brings up the whole platform behind trusted HTTPS. Re-run to upgrade.

BYOK

Your Anthropic key drives the agents. No per-seat AI markup — you pay your provider directly.

Read-only, non-destructive

The agents observe and diagnose, never mutating your code, data, or infra. Per-org isolation, restricted egress, encrypted secrets.

Built in

Everywhere your team
already works.

In your editor

An MCP server puts the agents inside Cursor and Claude Code — debug a flow without leaving your IDE.

In CI

A GitHub Action / GitLab CI and a CLI gate every PR/MR with a JUnit report and a real exit code.

In your tracker

The built-in board, plus GitHub, GitLab, or Jira — deduped by fingerprint.

In your infra

Read-only probes debug reachability, DNS, TLS, and dependency failures — not just UI bugs.

Pricing

Free to self-host.
A license for companies.

Source-available: run the whole platform yourself, bring your own key.

Self-hosted
Free

For any noncommercial use. Everything: chat debugging, backend root-cause, saved checks, CI gating. Bring your own LLM key.

Get started →
Commercial · for companies
Let's talk

For business & production use. A commercial license with terms that fit your team, plus deployment guidance.

Deploy it →

You ship it. The agents debug it.

Describe the bug in plain English — get a root-caused fix back. Self-hosted, your own key.