Hermes Train measures how you actually play — crosshair placement, counter-strafing, spray discipline, prefire angles — and builds every session around the skill you're worst at. No vibes. Just data.
A plugin inside the game server reads the engine's own state every time you pull the trigger. Not a stopwatch. Not a kill counter. The physics.
The angle between your crosshair and the enemy's head at the exact moment you fired. Good pre-aim means you're already on target when the angle opens — the difference between a flick and a micro-correction.
Your velocity at the fire tick. If you shoot while still sliding, the game adds inaccuracy — the engine penalises you and so does the score. Train the stop, not just the shot.
Burst length and recoil index per bullet. The system knows when you should have reset your spray instead of dragging bullet 20 into the ceiling.
Ask for a session and the coach reads your skill ratings and builds one with a reason behind every block. Try it:
The real engine picks the focus from your measured weakest skill — this is the same structure it generates.
Speed and accuracy are coupled in the score. Finishing 40% faster at half the accuracy scores worse, always — there's a regression test that proves it.
| Run | Time | Accuracy | Score |
|---|---|---|---|
| Sprint | 0:42 | 38% | 54 |
| Controlled | 0:58 | 81% | 88 |
Same drill, same bots — only the discipline differs.
Prefire routes for all nine Active Duty maps, three disciplines, three difficulty tiers — 55 scenarios per map, 6,315 bot placements in total.
Enemies placed at every common position along real routes. Clear the site like you're actually executing — because you are.
One-shot scenarios that punish sloppy crosshair placement and reward the first bullet landing where it should.
Buried bots with doubled HP force full-spray transfers — first ten bullets on point, resets where they belong.
From "start me a session" to measured progress — the whole chain is automated.