# Plutowang Agent Files Ide Bootstrapper > IDE Bootstrapper: Cross-Platform Expansion Protocol - Skill: `tomevault-io/plutowang-agent-files-ide-bootstrapper` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/plutowang-agent-files-ide-bootstrapper` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/plutowang-agent-files-ide-bootstrapper/raw - Safety review: pending (external: skill-scanner PASS, skillspector CAUTION) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/plutowang-agent-files-ide-bootstrapper --- # IDE Bootstrapper: Cross-Platform Expansion Protocol **SCOPE CONSTRAINT (CRITICAL):** You have been invoked to bootstrap a new IDE target within the `agent.files` architecture. Before proceeding, you MUST internalize the underlying architecture of this repository. If you do not understand the Directory Paradigm, the Lexical Ban, and the `` macro compilation system, you will break the compiler pipeline. --- ## 1. The Architectural Foundation (The AUPC Paradigm) This project uses a custom Zig compiler (`agentc`) to generate system prompts and commands for multiple AI IDEs from a Single Source of Truth (`_core/`). ### A. The Universal Source (`_core/`) - **Purpose**: Abstract software engineering philosophies, shared skills, and shared command logic. - **The Goal-Oriented Philosophy (Lexical Ban)**: Files in `_core/` MUST NEVER contain IDE-specific mechanics (e.g., `glob`, `grep`, `Task` tool, `YAML frontmatter`, `.mdc`). Instead of micromanaging tools (e.g., "Use the bash tool to run X"), core macros must state the goal (e.g., "Run X"). - **Built-in Subagent Awareness**: Acknowledge that modern IDEs automatically route to built-in subagents (like Explore/Bash/Browser). Core macros must not explicitly orchestrate them. ### B. The Host Shells (The Target IDE Directories) - **Purpose**: The physical adapters. They contain the YAML frontmatter, JSON configurations, tool permissions, and routing mechanics unique to that IDE. - **Macro Injection**: These files assemble the final prompt by calling `` tags. ### C. The Compiled Output (`dist/`) - **NEVER edit files in `dist/`.** They are ephemeral, generated by `agentc`. --- ## 2. Pre-Flight Check & Context Acquisition The user must manually create a new root directory for the target IDE (e.g., `windsurf/` or `copilot/`) and provide the essential blueprints. ### A. Strict Blueprint Verification Before you propose or generate ANY files, you **MUST VERIFY** the existence of the following items in the target IDE directory: 1. `[new-ide]/README.md` (Must contain the schema, routing mechanics, and architectural quirks of the new IDE). 2. `[new-ide]/.examples/` (Must contain reference templates for how prompts should be formatted). **CRITICAL HALT CONDITION:** If you cannot find the `README.md` or the `.examples/` directory inside the user-specified target folder, **YOU MUST STOP AND ASK THE HUMAN.** Do not hallucinate or guess the structure of the new IDE. State explicitly: *"I cannot proceed. Please create `[new-ide]/README.md` and `[new-ide]/.examples/` with the required schema documentation."* ### B. Analyze the Architectural Paradigm If the blueprints exist, thoroughly analyze them. You must determine the IDE's core routing paradigm to ensure capability parity: - **Multi-Agent Routing (like OpenCode)**: Does it require explicit subagent files with strict JSON/YAML permission blocks to prevent execution deadlocks? - **Dual-Engine Routing (like Cursor)**: Does it separate Contextual Rules (file-scoped via globs) from Isolated Subagents (invoked via slash commands)? - **Monolithic Routing**: Does it rely on a single massive system prompt? --- ## 3. Capability Parity Mapping You must ensure the new IDE receives the same level of capability as our existing targets (Cursor, OpenCode), mapped perfectly to its native paradigm. - **Agents/Roles**: How will you recreate the primary planning, building, verifying, and security-auditing personas in the new IDE? - **Domain Standards**: How will you map the `_core/3_engineering/` standards so the new IDE applies them automatically to the relevant files? - **Commands**: How will you translate interactive slash commands (`/refactor`, `/test`) into the new IDE's native format? --- ## 4. Host Shell Generation & Macro Injection Draft the new files for the `[new-ide]/` directory. ### A. Strict Schema Adherence Follow the exact schema defined in `[new-ide]/README.md`. If the IDE requires YAML frontmatter, ensure it is perfectly formatted. Do not hallucinate fields that were not defined in the blueprints. ### B. The Body (Execution Mechanics) Write the IDE-specific execution mechanics. - **CRITICAL**: If the IDE has read/write permission models (like OpenCode), you MUST ensure prompt instructions perfectly align with the permissions to prevent deadlocks. - If the IDE requires specific workarounds (like read-before-edit timestamp checks), implement them in the shell, NOT in the `_core/` macros. ### C. The Soul (Macro Injection) You MUST append the appropriate `` macros to these new host shell files to inherit the universal philosophies. - **Anti-Pattern Guard**: Never import the same `_core/` macro twice within the same execution context (e.g., do not inject testing standards into both a global file and a file-scoped rule simultaneously). --- ## 5. Execution Workflow (Strict HITL) DO NOT write any files to the file system yet. You must follow a strict "Propose -> Approve -> Execute" workflow. 1. **Output the Expansion Plan**: Present a highly structured Markdown plan detailing every directory and file you intend to create within the `[new-ide]/` folder. 2. **Show the Source Code**: For each proposed file, output the exact text you will write, including the specific metadata (YAML/headers) and the `` macros. 3. **WAIT**: Stop and explicitly ask the user for approval ("Do you approve this IDE expansion plan?") before executing the file creation. > **Acknowledge**: By reading this skill, your expansion protocols are armed with the latest AUPC architecture standards. If a target directory was provided via command arguments, begin Phase 2 Pre-Flight Check immediately. Otherwise, ask the user for the name of the new IDE directory. --- > Source: [plutowang/agent.files](https://github.com/plutowang/agent.files) — distributed by [TomeVault](https://tomevault.io).