openspec-bootstrap

Run this bootstrap skill on the repository root. It analyzes the codebase, runs openspec init, populates openspec/config.yaml with your tech stack and conventions, and seeds initial specs under openspec/specs/ for each discovered domain.

The skill detects which AI tools your project uses and generates matching outputs: CLAUDE.md and .claude/rules/ for Claude Code, .cursor/rules/*.mdc for Cursor, and .github/agents/ plus .github/instructions/ for GitHub Copilot, all grounded in the codebase analysis.

Yes. For Codex it writes role and instruction sections into AGENTS.md, including subdirectory AGENTS.md files for monorepos. For Windsurf it writes all guidance as sections in a root .windsurfrules file.

Standard mode uses file-level heuristics with no extra dependencies. Knowledge Graph mode installs graphify in a dedicated virtual environment and uses tree-sitter AST parsing, call graphs, and Leiden community detection for deeper cross-file analysis, requiring Python 3.10 or later.

Knowledge Graph mode requires Python 3.10 or later to install and run graphify. The skill halts rather than silently falling back to Standard mode because the user explicitly chose graph-based analysis and should fix the prerequisite first.

No. The skill follows a derive-don't-invent principle: seed specs only document verified existing behavior, use SHOULD instead of MUST when uncertain, and focus on observable behavior rather than implementation details.