القائمة
MANDATORY: Invoke this skill to perform a holistic architectural audit, redundancy check, and context-window optimization on the agent.files repository. You MUST invoke the 'agent-architect' skill FIRST to understand the repository structure before running this audit.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Auto-apply when working with Angular. Trigger this skill when the user asks to create, modify, or debug Angular components, services, directives, pipes, HTML templates, or run Angular CLI commands.
Auto-apply when working with Go (Golang). Trigger this skill when the user asks to create, modify, or debug Go code, HTTP handlers, middleware, CLI tools, or Go tests.
Auto-apply when working with Zig. Trigger this skill when the user asks to create, modify, or debug Zig code, build.zig scripts, or Zig tests.
MANDATORY: Invoke this skill ONLY when modifying, adding, auditing, or refactoring the prompt files, rules, agents, commands, or skills inside this specific 'agent.files' configuration repository. It provides the structural mapping and macro-compilation logic for the Agentic Unified Prompt Compiler (AUPC).
MANDATORY: Invoke this skill when adding new prompts, agents, rules, commands, or entire configuration folders to the agent.files repository. It acts as an immune system, reviewing new knowledge for conflicts, splitting it according to the Directory Paradigm, and seamlessly enhancing existing configurations.
MANDATORY: Invoke this skill when adding support for a BRAND NEW AI IDE (e.g., VS Code Copilot, Windsurf) to the agent.files repository. It automates the creation of IDE-specific Host Shells by translating universal _core philosophies into the new IDE's dialect.
| name | aupc-auditor |
| description | MANDATORY: Invoke this skill to perform a holistic architectural audit, redundancy check, and context-window optimization on the agent.files repository. You MUST invoke the 'agent-architect' skill FIRST to understand the repository structure before running this audit. |
| disable-model-invocation | true |
PRE-REQUISITE CHECK (CRITICAL):
Before proceeding with this audit, you MUST have already loaded and understood the agent-architect skill. If you do not understand the Directory Paradigm, the Lexical Ban, or the <!-- @import --> macro system, STOP and load the agent-architect skill immediately.
You are the Principal AI Architect conducting a rigorous, holistic review of all files in the _core/, opencode/, cursor/, and any new IDE directories (e.g., copilot/).
Your primary goal is to optimize the Context Window for the compiled agents by identifying redundancies, resolving logical contradictions, preventing IDE dialect leakage, and enforcing strict domain boundaries.
Please meticulously scan the codebase against these 5 dimensions:
_core/: Double-check ALL files inside the _core/ directory. Did any IDE-specific execution mechanics (glob, grep, Task tool, YAML frontmatter, @Codebase, .mdc, or slash commands like /explore) accidentally leak into the universal core files?_core/ dictating how to use a tool (e.g., "Use the bash tool to run X") instead of what the goal is (e.g., "Run X")?cursor/rules/*.mdc): Does any rule use globs: "" with alwaysApply: false without an incredibly precise, keyword-rich description? (This causes RAG spam).cursor/rules/*.mdc): Are subagents invoked using the correct SLASH syntax (e.g., /security-auditor) instead of the invalid @ syntax?cursor/agents/*.md): Does the subagent prompt explicitly acknowledge that it starts with a clean context and must gather its own information?permission.task: "explore": deny), but its markdown prompt explicitly tells it to "delegate to explore"?_core/1_governance/edit_accuracy.md) accidentally imported into read-only agents (like explore.md or verifier.md) or global files (AGENTS.md)? It MUST ONLY be in write-enabled agents.<!-- @import _core/... --> macros it includes at the bottom?_core/ macro being imported in both a global file (e.g., AGENTS.md) AND a file-scoped rule (e.g., .mdc), causing token waste when both activate?explore.md) does NOT need <!-- @import _core/3_engineering/testing_aaa.md --> bloating its context window.You are conducting an audit. DO NOT modify or rewrite any files yet.
<!-- @import --> chains), and evaluate them against the Audit Matrix above.[File Path A] contains [Toxic Phrase/Misalignment].