@francisfuzz's dotfiles
This repository uses a consolidated configuration approach with .agents/ as the single source of truth, accessible to both Claude Code and GitHub Copilot CLI via symlinks.
- agents/ – Subagent definitions for delegation (explore, librarian, metis, momus, oracle)
- skills/ – Reusable skills invoked via
/skill-name:git-ops– Atomic commits, interactive rebase, history searchinterview– Discovery interviews with Socratic questioningreview-work– Post-implementation review with parallel agentstranscript-to-artifact– Meeting transcripts → structured artifacts
Previously active configuration that has been retired:
- commands/ – Legacy Claude command templates
- prompts/ – Legacy prompt templates
- skills/ – Archived skill definitions (git-commit, pr-review-assist, etc.)
| Symlink | Target | Purpose |
|---|---|---|
CLAUDE.md |
AGENTS.md |
Both tools read the same agent config |
.claude/agents |
.agents/agents |
Claude Code subagent discovery |
.claude/commands |
archive/commands |
Legacy command compatibility |
.claude/skills |
.agents/skills |
Claude Code skill discovery |
.github/prompts |
archive/prompts |
Legacy prompt compatibility |
.github/skills |
.agents/skills |
Copilot CLI skill discovery |
GNU General Public License v3.0
- Interview Skill Inspiration:
@developersdigest"Claude Code 'Interview' Mode in 6 Minutes" - Agent & Command Patterns:
francisfuzz/learning-opencode