Doksi

Your AI agent writes code, fixes bugs, ships PRs. Then you ask it to check if the login screen actually works on mobile and it can't. So you open the emulator. Tap around. Copy the error. Paste it back. The agent fixes it. You test again. Manually. Again. We built Doksi Device MCP to kill that loop. Your agent gets a cloud device with 50+ tools. Not just screenshots and taps. The full picture: → Every HTTP request and response body your app makes → Structured app logs (not a wall of raw logcat) → Internal app data, databases, preferences, files → System events, SMS, phone calls, GPS, biometrics, network conditions The agent installs the build, navigates the app, reads what's happening underneath the UI, and tells you what's broken. You never leave your editor. Works with Claude Code, Codex, Cursor, Copilot, Windsurf. Test your cross-platform mobile apps right where you code. Here's Claude Code finding (and later fixing!) bugs using Doksi Device MCP. Two bugs invisible on screen, both caught.

Today we are launching "Schedules" on Doksi. Just tell Doksi how often you want to run a test and it handles the rest. Doksi will tell you when things break, before your customers do. Signup here 👉 app.doksi.ai/signup

Today we're launching Doksi for E2E testing. 🚀 Describe your tests in plain English. Doksi handles the rest. Here's what it can do: → Test any Mobile app. No SDK needed → Navigate multi-app flows autonomously (app switches, email verification, OAuth, OTPs) → Self-healing. UI changes don't break your tests → Full network visibility. Every request and response, headers and bodies, even on production apps → Device-level debugging. Logs, CPU, memory, all captured → CI/CD native. Trigger from GitHub Actions, GitLab, Slack, or on a schedule → AI-generated test reports that tell you not just what broke, but why. Here's a 2-min demo testing Reddit's forgot password flow across two apps — fully autonomous. Sign up at https://lnkd.in/eDiFQwHA