Perception and suspicion
See, hear, smell, discover clues, respond to lighting and crouching, and escalate through five readable suspicion levels.
Build enemies that act on evidence—not hidden player state.
A modular Unity 6 runtime that connects perception, suspicion, locomotion, memory, decisions, squad tactics, cover combat, health, aiming, and weapons into one explainable stack.
Argus Pro is being built around observations, timestamps, confidence, reports, and explicit contracts. Enemies can infer. They can be wrong. They can receive old information. They can coordinate without secretly knowing where the player is.
Use the complete Drop-In enemy or integrate only the modules and interfaces your game needs.
See, hear, smell, discover clues, respond to lighting and crouching, and escalate through five readable suspicion levels.
Keep useful observations after contact is lost instead of collapsing every decision into the current frame.
Exchange timestamped reports with nearby allies while preserving age, uncertainty, and the difference between seeing and hearing.
Choose actions through readable utility evidence and contracts—not privileged access to hidden player state.
Divide search and investigation work so a squad covers useful space instead of checking the same point repeatedly.
Predict likely escape trajectories from genuine observations, then assign pursuit, interception, and containment work.
Claim cover, peek in bounded windows, suppress, advance, hand off covering fire, reposition, and retreat when wounded.
Read scoring evidence, reasoning traces, contracts, assignments, and debugging signals while the encounter is running.
Every step preserves the evidence that caused the next one.
Create evidence from senses and world events.
Retain timestamped observations after contact breaks.
Pass imperfect reports to allies without manufacturing certainty.
Compare actions against current evidence, needs, and contracts.
Assign non-overlapping roles, positions, and tactical work.
Move, search, take cover, fight, recover, or retreat.
The target is not six agents sprinting toward one last-known point. Pro assigns distinct tactical work and keeps positions claimed so coordination survives contact with the level.
Discover cover or use authored points. Claim separate positions. Peek for bounded firing windows. Suppress while one teammate advances. Hand off covering fire. Recover, reposition, and retreat when wounds change the value of the fight.
Pro contains Core. They are alternative installs, not packages that should be stacked together.
Add Argus Pro Enemy (Drop-In) to assemble the full reference stack, or connect
individual modules through public C# interfaces. The included NavMesh implementation is a
starting point, not a lock-in strategy.
IArgusProLocomotion Keep cognition and tactics. Swap the component that moves the body.
See the complete stack operating in an encounter rather than isolated editor screenshots.
Learn one system at a time with scenes built around a single mechanic or integration boundary.
Exercise the Pro systems through fifteen deliberately scoped tactical situations.
Read, modify, debug, and own the runtime. No black-box DLL and no subscription.
Bring your own production character art and animations. Argus Pro is deterministic gameplay AI: no machine learning, no generative AI, no cloud service, no telemetry dependency, and no subscription.
No. Argus Pro includes the complete Argus Core feature set. Install Core or Pro—never both—in the same project.
Core provides perception, awareness, suspicion, basic guard behavior, and alert propagation. Pro adds durable memory, imperfect intelligence sharing, utility decisions, coordinated investigation and search, predictive interception, tactical actions, combat formations, cover combat, fire-team coordination, advanced tuning, and reasoning tools.
No. Prediction and tactical decisions are built from genuine observations and reports. The design goal is enemies that reason from evidence rather than omniscient access to hidden player state.
No. The built-in NavMesh locomotion and reference combat stack provide a complete starting path, but public C# interfaces are designed for custom locomotion, root motion, KCC, A*, ECS, another controller, or your own production combat systems.
Bring your own production character art and animations. Argus Pro provides enemy cognition, tactical decisions, integration contracts, and a reference combat stack.
No. Argus Pro uses deterministic gameplay AI. It has no machine-learning model, generative AI service, cloud dependency, or subscription.
It is still in active development, and the feature set and presentation may change before release. This page documents the current direction without promising a release date.
Now the enemy gets memory, judgment, teamwork, and a reason for every move.