pcx-ai-toolkit

Techniques for driving AI coding tools (Claude Code, Cursor, Cline, Aider, Copilot) effectively on PCX projects. Covers context loading, prompt discipline, and steering the AI to read docs before writing API calls. Always active when working with an AI on a PCX scripting project.

Mandatory doc router for all PCX scripting sessions. Triggers on any game hacking, game cheat, ESP, aimbot, triggerbot, radar, Enma, AngelScript, or Perception.cx work. Provides the full supported doc index (32,000+ lines across 123 docs) and enforces reading the relevant documentation before writing any API call. Load alongside game-cheat-script-master and game-cheat-guidelines on every PCX game-cheat session.

Runs the PCX AI Toolkit CLI directly from Claude Code. Use this for `/pcx update` or any other pcx command.

Use for Source 2 live MCP investigations with schema/interface evidence.

Use for Unity IL2CPP live MCP investigations with GameAssembly and global-metadata evidence.

Use for Unreal live MCP investigations with UObject/GName/vtable evidence.

Use when recording, verifying, or rendering machine-readable reverse-engineering claim/evidence graphs.

Routing guide for the 59 Perception MCP tools — which to pick for memory reads, scans, disassembly, PE/module walks, and the Enma scripting bridge to avoid slower or redundant calls. Always active when calling Perception MCP tools. Answers "which tool for this task" so the AI picks the cheapest tool with the required precision.

Use for reverse-engineering investigations that need Perception MCP workflow chains and claim/evidence proof.

Use before Perception MCP write, allocation, kernel, or broad scan operations to enforce permissions and rollback.

Use for live Perception MCP sessions: acquiring handles, making calls, retrying stale handles once, recording transcripts, and cleaning up.

Mandatory behavioral rules and practical patterns for writing Perception.cx game-cheat scripts in Enma. Always active — these rules apply every time you write or edit game-cheat code, including ESP, aimbot, triggerbot, radar, pattern scanning, and overlay rendering. Authorized use only — analyze software you own or are permitted to test.

Master skill for Perception.cx game cheat script development. Activates on any request for cheats, hacks, ESP, aimbot, triggerbot, radar, overlay, memory reading, pattern scanning, or anti-cheat research. Loads the complete doc index, the 12 cheat guidelines, the AI pair-programming workflow, and the official cheat templates. Always use this alongside game-cheat-guidelines and game-hacking-pcx.

Behavioral and syntactic rules for writing .em (Enma) scripts on Perception.cx. Prevents AngelScript-reflex errors in the Enma API surface — method names, parameter shapes, type system, and lifecycle differ from AS. Always active when editing .em files.

Guide to the three surfaces (llms.txt static index, bundle files, MCP server) through which AI tools reach the toolkit corpus, and which to pick under which integration model. Always active when working with the pcx-ai-toolkit knowledge base from any AI tool (Claude Code, Cursor, Cline, Aider, Copilot, Continue, Zed).

Methodology for reverse engineering kernel-level anti-cheat systems: EAC, BattlEye, Vanguard, GameGuard, XIGNCODE3. Covers component enumeration, detection-vector cataloging, and verified understanding of the AC observation surface. Always active when analyzing anti-cheat systems.

Methodology for reversing binaries protected by commercial obfuscators (Themida, VMProtect), compiler-level obfuscation (OLLVM, Hikari), and custom protection schemes. Covers identification, classification, layer stripping, devirtualization, and verification. Active when analyzing obfuscated or packed binaries.

Technical patterns for reversing Windows kernel drivers: WDM/KMDF structure identification, IOCTL dispatch tables, kernel callback enumeration, integrity checks, obfuscation layers, and driver communication protocols. Focused on anti-cheat driver analysis. Always active when analyzing kernel driver binaries.

Pattern for supporting multiple game binaries (versions, architectures, storefronts, beta channels) from one Enma codebase without forking. Triggers when the user mentions multiple game versions, cross-build support, 32-bit vs 64-bit, multi-store builds, or maintaining stable and experimental branches in parallel.

Workflow discipline for reverse engineering and offset maintenance: locating structs, generating signatures, resolving RIP-relative addresses, and keeping an offset table alive across patches. Derived from Karpathy principles, rewritten for RE where the failure mode is a confident wrong answer. Always active when doing RE or offset work.

Discipline for recording why each offset and sig is trusted — the proof behind the offset table. Every offset added, every sig derived, every struct layout committed comes with a citable evidence entry. Always active during RE work; pairs with pcx-re-discipline and pcx-patch-day-playbook.

Guidelines for developing hybrid Rust-Python tools in pcx-ai-toolkit. Covers the multi-binary Cargo architecture, transparent Python proxying with fallbacks, match ergonomics, and zero-dependency mock PE testing.

Build and deployment workflow for PCX scripts: .em vs .emb, bundle order respecting the module-import graph, hot-reload survival, pre-ship hygiene checklist, runtime-version pinning, and distribution metadata. Triggers when packaging, distributing, or releasing scripts to other users.

Guides the AI and developer to build small, modular verification scripts or test routines that read target memory and validate struct layouts, offsets, and type shapes before writing complex rendering or logic code.

Workflow discipline for developing Enma (.em) and AngelScript (.as) scripts on Perception.cx. Derived from Karpathy principles — think before coding, simplicity first, surgical changes, goal-driven execution — rewritten for cheat development realities: stale offsets, silent failed reads, detection surface. Always active when writing or editing PCX scripts.

Pattern for shipping diagnostic and profiler output as a separate, gated overlay rather than mixing it into the production rendering. Triggers when debugging a script, building a support-mode panel, profiling a slow path, or creating a diagnostic vs release build of the same code.

Ordered triage workflow for recovering a PCX script after a game update. Triggers when sigs return 0, reads return garbage after a patch, or the user says "broken", "updated", "patch day", "hotfix", "season drop", or "DLC dropped". Keeps diagnosis short and fixes targeted.

Turns the update/render separation rule into enforceable numeric budgets using mono_us() measurements. Covers per-frame targets at common refresh rates, per-call cost rules of thumb, a drop-in profiler recipe, and read-coalescing patterns. Always active when writing or reviewing performance-sensitive render or update routines.

Explains when PCX overlay output appears in screen captures per capture method (OBS, Discord, ShadowPlay, NVIDIA Highlights, capture cards, PrintScreen). Triggers on streaming, OBS, Discord screenshare, "my overlay shows on stream," "my friend can see my menu," and related capture or recording questions.

Whole-project audit for over-engineering in PCX scripts. Scans every .em and .as file for wrappers, dead abstractions, duplicate entity walks, unused offsets, and config systems that outweigh their settings. Ranked by lines recoverable. Use when the user says "audit for bloat", "what can I delete", "find over-engineering", "slim this project", or invokes /pcx-bloat-audit. One-shot report, does not apply fixes.

Code review focused on over-engineering in PCX scripts. Finds what to delete: proc_t wrappers, entity managers for three entities, config systems for two settings, class hierarchies for one feature. One line per finding. Use when the user says "review for bloat", "is this over-engineered", "what can I delete", or invokes /pcx-bloat-review. Complements correctness review — this one only hunts unnecessary complexity in Enma/AngelScript scripts.

Harvest every `// defer:` comment in the PCX project into a debt ledger. Tracks deliberate shortcuts (global handles, hardcoded colors, single-walk assumptions) so deferrals don't rot into permanent hacks. Use when the user says "defer ledger", "what did we defer", "list shortcuts", "show debt", or invokes /pcx-defer-ledger. One-shot report, changes nothing.