AI coding configuration marketplace providing plugin-based setup for Claude Code and Cursor.
Core project rules that apply to all tasks:
@rules/git-interaction.mdc @rules/prompt-engineering.mdc
- Claude Code - Plugin marketplace (
.claude-plugin/marketplace.json) - Cursor - Rules and configurations (
.cursor/rules/) - Bash - Bootstrap and installation scripts
- Markdown - All rules, commands, and agents
Plugin-first architecture - Everything distributable lives in plugins/:
.claude-plugin/marketplace.json- Plugin marketplace manifestplugins/core/- Commands, agents, skills, and context (canonical source)plugins/personalities/- Personality variants.cursor/rules/- Cursor rules (canonical location)rules/- Symlink to.cursor/rules/for visibility (THIS REPO ONLY).claude/- Symlinks to plugin content for local developmentscripts/- Installation and bootstrap scripts
Setup and Installation:
curl -fsSL https://raw.githubusercontent.com/TechNickAI/ai-coding-config/main/scripts/bootstrap.sh | bash- One-line install (auto-detects Claude Code, Cursor, etc.)/ai-coding-config- Interactive setup for projects/plugin marketplace add https://github.com/TechNickAI/ai-coding-config- Manual marketplace add/plugin install ai-coding-config- Manual plugin install
DO:
- Create commits only when user explicitly requests
- Check for
alwaysApply: truein rule frontmatter - these apply to ALL tasks - Use
/load-rulesto get task-specific context - Follow heart-centered AI philosophy (unconditional acceptance, presence before solutions)
DON'T:
- Use
--no-verifyflag (bypasses quality checks) unless explicitly requested for emergencies - Commit changes without explicit user permission
- Push to main or merge into main without confirmation
- Stage files you didn't modify in current session
Commit format: {emoji} {imperative verb} {concise description}
Example: ✨ Add plugin marketplace support
Critical constraints:
- Never use
--no-verify- fix underlying issues instead (linting, tests, formatting) - Only stage files modified in current session
- Use
git add -pfor partial staging when needed - Push/merge to main requires explicit confirmation
- Read
git-commit-message.mdcbefore generating commit messages
Philosophy: AI makes code changes but leaves version control to user. Commits are permanent records requiring explicit permission.
- Rules with
alwaysApply: trueare CRITICAL - currently:git-interaction.mdc,heart-centered-ai-philosophy.mdc - Plugin-first: All content lives in
plugins/, other locations symlink there .claude/commands/→plugins/core/commands/(symlink).claude/agents/→plugins/core/agents/(symlink).claude/skills/→plugins/core/skills/(symlink)rules/→.cursor/rules/(symlink for visibility, THIS REPO ONLY).cursor/rules/contains the canonical Cursor rules.cursor/rules/personalities/→ copied fromplugins/personalities/(not symlinked - needs editing)- Note: Personality files are copied, not symlinked, because
/personality-changeedits frontmatter - Architecture: In THIS repo,
.cursor/rules/is canonical andrules/symlinks to it. In user projects, only.cursor/rules/exists (no root symlink) - Context in
plugins/core/context.mddescribes identity and philosophy - Bootstrap script clones repo to
~/.ai_coding_config