Reference
Troubleshooting
Installation and activation
| Symptom | Likely cause and fix |
|---|---|
| The installer stops on the activation page | The Activation key field is blank or wrong. Paste the full sk-pharmalytics-… key with no spaces. |
| "Invalid or inactive API key" | Mistyped, or the key has been rotated. Rotate it in Devices and use the fresh key. |
| "This activation key is already in use on another computer" | One key binds to one machine. Use Rotate API key on that device to clear the binding, then paste the new key. |
| "Could not reach the backend at…" | No internet, or HTTPS is blocked. Ask IT to allow outbound HTTPS to *.pharmalytics.skypondtech.com. |
| "The backend returned a server error" | Temporary service issue. Wait a minute and retry. |
| The agent asks for the activation key even though the installer took it | The activation landed in a different Windows profile — usually because the agent was launched from the admin account used to install. Sign in as the operator account and use Update API Key. |
| Identity panel shows dashes and "Update the API key." | The stored key is no longer valid. Click Update API Key and enter a current one. |
| Windows SmartScreen blocks the installer | Expected — the installer is distributed privately. More info → Run anyway. |
The agent is not processing
| Symptom | Likely cause and fix |
|---|---|
| Badge shows ● Paused | Almost always QS1 is closed or signed out. Open QS1, sign in, leave it open. The agent resumes on its own — do not press Start. |
| Badge shows ● Sleeping | Normal. The queue was empty; it will re-check at the time shown. |
| Badge shows ● Error | It stopped on a fault. Confirm DocuTrack and QS1 are open and signed in, then press Start. If it recurs with the same message, send Skypond the alert email. |
| Nothing has run since a reboot | This build registers no scheduled task. Someone must open PharmAutomate and click Start. |
| Nothing runs and the badge looks fine | Check Schedules — if every workflow for that branch is disabled, the agent has no window to work in. |
| "The screen cannot be read" | The workstation is locked or its RDP session is detached. Reconnect and leave the session attached; see Before you begin. |
| The DocuTrack folder is not found | Configure the full folder name. Cancel is ambiguous when Cancel with Association exists. |
| DocuTrack or QS1 does not come up when a run starts | The exe path is not set. Settings → Worker Settings → Exe path → Browse… for both. The agent still never signs in, and never restarts QS1. |
| The first Start is slow | One-time install of the Tesseract text-recognition component. Let it finish. |
| Every document goes to manual review | Tesseract may be missing and winget unavailable on that machine. Check the live log; install Tesseract and restart the agent. |
| The mouse moves on its own | Automation is running on that computer. Press Stop before using it, or dedicate a PC/VM to automation. |
| A second agent refuses to start | Deliberate — two agents would share one mouse. Use the control window that is already open. |
Portal and access
| Symptom | Likely cause and fix |
|---|---|
| Set-password link expired | Links last 24 hours. Use Forgot password, or ask Skypond support for the administrator account. |
| A staff member cannot see a branch | Their location-viewer account is scoped per branch. Edit their branches in Users. |
| You need to change someone's role | Role changes are done by Skypond, not in the merchant portal. |
| Documents skipped or marked manual review | Normal. The agent only completes what it reads confidently; the rest is left for your team, never guessed. |
| A device shows as stale on the Dashboard | Last seen more than 7 days ago, or never. Check whether that machine is still in service. |
No alert arrived
See Why didn't I get an email? — the Recent alerts card records every alert considered, including the ones that were not sent and why.
What to send support
- The device name / Agent ID and branch.
- The approximate time.
- The alert email, if you got one.
- Ideally, a copy of the device's live log while the problem is happening — the Copy button on the device page.