Integrated Development Environment API Rules

aid: integrated-development-environment:rules
name: Integrated Development Environment Usage Rules
description: >-
  Operational rules and guardrails for evaluating, configuring, and operating
  Integrated Development Environments (IDEs). These rules cover workspace
  setup, source control, language server use, AI assistant safety, secrets
  handling, and remote development practices that apply across editors and
  vendor ecosystems.
modified: '2026-04-28'
rules:
  - id: workspace-isolation
    category: Workspace
    statement: >-
      Open each project in its own workspace or window so that language server,
      task, and debugger configuration do not leak between unrelated codebases.
  - id: lsp-per-language
    category: Language Servers
    statement: >-
      Prefer the official language server (LSP) for each language and pin its
      version in repository configuration to keep diagnostics and refactors
      consistent across machines.
  - id: source-control-required
    category: Source Control
    statement: >-
      Require a clean source control state before running large refactors,
      automated edits, or AI-driven code generation so that changes can be
      reviewed and reverted as a single unit.
  - id: secrets-out-of-workspace
    category: Security
    statement: >-
      Keep API tokens, cloud credentials, and signing keys in OS keychains or
      .env files excluded from source control rather than in checked-in IDE
      settings.
  - id: extension-trust
    category: Extensions
    statement: >-
      Install editor extensions only from the vendor marketplace or a verified
      publisher and review the requested permissions before enabling.
  - id: ai-assistant-scope
    category: AI Coding
    statement: >-
      Restrict AI coding assistants to the active workspace and explicitly
      whitelisted directories; do not grant blanket access to the home
      directory or system-level paths.
  - id: ai-assistant-review
    category: AI Coding
    statement: >-
      Review every AI-suggested edit through the diff viewer before accepting,
      and run the project's tests or type checker before committing.
  - id: mcp-server-allowlist
    category: AI Coding
    statement: >-
      When connecting Model Context Protocol (MCP) servers to an editor,
      maintain an allowlist of servers and tools, and prefer read-only or
      sandboxed servers for untrusted projects.
  - id: remote-dev-isolation
    category: Remote Development
    statement: >-
      Use dev containers, codespaces, or SSH remotes to isolate project
      toolchains rather than installing every language runtime globally on the
      workstation.
  - id: format-on-save-pinned
    category: Code Quality
    statement: >-
      Pin formatter and linter versions per repository (e.g., .prettierrc,
      .editorconfig, ruff/eslint config) so that format-on-save behaves
      identically across contributors and IDEs.
  - id: debugger-source-maps
    category: Debugging
    statement: >-
      Generate and ship source maps for compiled and bundled languages so that
      breakpoints, stack traces, and step-through debugging map back to the
      original source.
  - id: telemetry-policy
    category: Privacy
    statement: >-
      Document and enforce a team telemetry policy for the IDE and its
      extensions, disabling telemetry that is not required when working with
      regulated or proprietary code.