Stop typing alliance leaderboards by hand.

The scanner supports the in-game event, roster, participant-mail, and result screens that officers otherwise type into Discord or spreadsheets:

• Marshal's Guard — scores and damage per member.
• Sandworm Battle — scores and damage dealt per member.
• Alliance Duel — daily and weekly rankings, attendance, and per-day VS scores.
• Tech Donations — donations per member, the VS Day 6 score-based leaderboard.
• Kills — kill counts per member.
• Zombie Siege — waves cleared and damage dealt per member.
• Member lists — bulk-create placeholder member profiles from an in-game alliance roster screenshot.
• Hero Power lists — per-member power readings from the hero roster screen.
• Desert Storm rosters — sign-up screens, so the platform knows who said they'd play before the match.
• Desert Storm participant mail — confirmed participant names and power from the in-game message.
• Desert Storm results — post-match scoring, MVP awards, and per-player individual scores.
• Canyon Storm rosters and results — starters, substitutes, post-match scoring, and individual scores.

Each one is available in its matching form. The Screenshot Inbox can also classify a mixed batch and route every result to the right review flow before anything is saved. See the inbox workflow.

Accuracy on a clean in-game capture is typically 95%+ on player names and 99%+ on numbers. The model is tuned on raw in-game screen captures, so it understands the in-game UI's column layouts, the suffix conventions on big numbers (M, B, T), and the rank badges next to player names.

The two failure modes worth knowing about: unusual characters in player names (the in-game font is tight on accented characters and emoji), and very low-resolution captures from older devices. In both cases the platform falls back to a review screen where the officer can correct any row before save — there's no “set and forget” mode that could write the wrong number.

For a mixed set, open Screenshot Inbox and add up to 10 screens. For one known workflow, open its ranking or result form and use the camera control there. Multiple screenshots are fine — for a long leaderboard, take captures that overlap by a few rows and review the combined rows before saving.

Individual scanners accept up to 20 screenshots per extraction; the mixed inbox accepts 10 at a time. The model returns an extracted-rows preview within a few seconds. Officers review, edit any miscaptured row, and save. Members are matched to existing profiles using the tool's normal review logic, so confirmed scores feed the right member's history.

The full how-to and tips for getting good scans is in AI screenshot scanning.

Last Command never connects to the in-game APIs and never scrapes anything from your device — the scanner only reads screenshots an officer explicitly uploads. There's no mobile camera mode, no OCR overlay, and no automated capture of in-game screens. If you can paste it into Discord, you can paste it into Last Command.

Pro alliances get 300 extractions per month, with up to 20 screenshots per scan. That's well above what an active alliance uses even when scanning Marshal's Guard, Alliance Duel, donations, and Desert Storm results every cycle. The cap exists to keep cost predictable, not to meter normal use.

Screenshots go to AWS Bedrock for the extraction step (in the EU region) and aren't retained after the response comes back. The model isn't trained on your alliance data — we don't have a feedback loop that quietly improves the scanner from your screenshots. If we ever did, it would be opt-in with explicit notice.

The full security and data-handling story is in our Security page and Privacy policy.