Machine-readable architecture decision records. Git-native decision memory and governance for human and agent-authored plans -- lint, CI, and MCP.
Latest release v0.20.0 · 11 Oct 2026
15 files for GitHub Copilot, Claude Code, Codex and OpenCode: adrkit copilot-instructions.md, adrkit use-bun.instructions.md, code-review, adrkit AGENTS.md and 11 more — 24,073 tokens loaded in every session.
.github/copilot-instructions.md A 310 tok .github/instructions/use-bun.instructions.md A 776 tok AGENTS.md A 22,724 tok CLAUDE.md A 263 tok .github/skills/code-review/SKILL.md A 48 tok .github/prompts/speckit.analyze.prompt.md A — .github/prompts/speckit.checklist.prompt.md A — .github/prompts/speckit.clarify.prompt.md A — .github/prompts/speckit.constitution.prompt.md A — .github/prompts/speckit.converge.prompt.md A — .github/prompts/speckit.implement.prompt.md A — .github/prompts/speckit.plan.prompt.md A — .github/prompts/speckit.specify.prompt.md A — .github/prompts/speckit.tasks.prompt.md A — .github/prompts/speckit.taskstoissues.prompt.md A — These files are mbeacom/adrkit's own configuration — they tell GitHub Copilot, Claude Code, Codex and OpenCode how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.