At MARS 2614, we don’t just write code; we teach AI how to write code for us. This is the heart of technical Innovation. MARSLib is fundamentally co-developed with Artificial Intelligence models, which act as active participants in our journey of Discovery.
Beyond code generation, our AI is designed to act as a Socratic educational mentor through Teamwork. We believe in using AI to ensure Inclusion—empowering every student to deeply learn the complex robotics engineering and physics principles underlying our framework, making the mastery of advanced technology a Fun and engaging experience.
| Agent Skill Name | Purpose & Workflow |
|---|
marslib | The general-purpose framework architect. Use when adding new sub-systems or managing generic IO patterns. |
marslib-core-standards | Enforces the “Never Nester” rule, explicit mathematical documentation, and strict structural aesthetics on modified classes. |
marslib-audit | Executes a championship-grade code audit across the repository to enforce maximum reliability. |
marslib-skill-authoring | Helps author and serialize brand new .agent skills to scale the framework’s intelligence. |
marslib-tutor | Overrides standard productivity directives to become a patient, Socratic mentor. Prioritizes teaching the “Why” underlying FRC physics and structure over just dumping code. |
| Agent Skill Name | Purpose & Workflow |
|---|
marslib-mechanisms | Scaffolds standard IO abstraction layers (RotaryMechanismIO, LinearMechanismIO) and test simulation layers. |
marslib-swerve | Extends and configures the standard SwerveDrive subsystem, 250Hz odometry logging, and PhoenixOdometryThread integration. |
marslib-superstructure | Orchestrates the MARSSuperstructure multi-state mechanical coordination layer. |
marslib-statemachine | Implements multi-state mechanisms, sequencing logic, and transition validation checks. |
marslib-operator | Configures Command Controller layouts, mapping triggers, and haptic rumble pipelines. |
| Agent Skill Name | Purpose & Workflow |
|---|
marslib-control-theory | Tunes PID loops, continuous feedforwards, and mechanical slew limiters. |
marslib-math | Processes vector transformations, regression filters, and geometric logic interpolation. |
marslib-shotsetup | Tunes the internal Shot-On-The-Move (SOTM) iterative target-convergence solvers and physics trajectories. |
marslib-vision | Integrates Megatag fusion, multi-camera pipelining, and AdvantageScope 3D field layouts. |
| Agent Skill Name | Purpose & Workflow |
|---|
marslib-simulation | Registers Dyn4j physics bodies, adds field obstacles, and regulates bounding box interactions. |
marslib-testing | Automates the writing of JUnit 5 tests against physical simulation to validate code prior to dropping onto physical hardware. |
marslib-telemetry | Manages AdvantageKit log replay integration, variable reporting, and replay-only execution loops. |
marslib-diagnostics | Injects pre-match system health sweeps and parses CAN bus dropout assertions. |
marslib-power | Fine-tunes hardware current limits, limits CAN bus utilization, and enacts brownout protection. |
| Agent Skill Name | Purpose & Workflow |
|---|
marslib-vscode-extension | Expands the standalone Visual Studio Code extension with new linters, trees, and dashboard commands. |
marslib-cloudflare | Deploys this very Starlight/Astro documentation hub to Cloudflare pages and reduce 404/SSR worker errors. |
marslib-web-accessibility | Plugs into the Pa11y tests to execute automated WCAG 2.1 AA audits on DOM structure. |
advantagescope-layouts | Governs the .json structure of customized AdvantageScope dashboards. |
Whenever an AI algorithm is tasked to structure a specific code loop (like the SOTM targeting loop or logging arrays), use the @AGENT_ENFORCED block tag right inside your code editor to explicitly warn human developers from accidentally refactoring crucial, mechanically generated syntax patterns.
