Skill: core-auto-detect

What I do

I detect project environments by scanning root-level files and recommend appropriate skills to load, enabling automatic domain expertise activation without explicit configuration.

When to use me

• Starting a new development session in an unfamiliar project
• Automating skill selection in CI/CD or batch workflows
• Reducing manual skill specification overhead

Core principles

1. File-presence detection — Check root directory only (no recursive scanning)
2. Skill mapping — Each environment maps to curated recommended skills
3. Non-invasive — Read-only, offline, millisecond completion
4. Composable — Multiple detections fire simultaneously

Detection rules & skill recommendations

Go Projects

Detection: go.mod exists

Recommended skills: golang, ginkgo-gomega, clean-code

Node.js / JavaScript Projects

Detection: package.json exists

Recommended skills: javascript, jest, clean-code

Ruby Projects

Detection: Gemfile exists

Recommended skills: ruby, rspec-testing, clean-code

Python Projects

Detection: pyproject.toml or setup.py exists

Recommended skills: python, clean-code

Embedded / Microcontroller Projects

Detection: platformio.ini exists

Recommended skills: cpp, platformio, embedded-testing

Rust Projects

Detection: Cargo.toml exists

Recommended skills: rust, clean-code

Nix / NixOS Projects

Detection: flake.nix or shell.nix exists

Recommended skills: nix, devops

CI/CD / GitHub Actions

Detection: .github/workflows/ directory exists

Recommended skills: github-expert, devops, automation

Build Automation

Detection: Makefile exists

Recommended skills: automation, scripter

Patterns & examples

Single-language: go.mod → golang, ginkgo-gomega, clean-code

Polyglot with CI/CD: go.mod + package.json + .github/workflows/ → golang, ginkgo-gomega, javascript, jest, github-expert, devops, automation, clean-code

Embedded with build: platformio.ini + Makefile → cpp, platformio, embedded-testing, automation, scripter

Anti-patterns to avoid

• ❌ Recursive scanning — Check root directory only
• ❌ Network calls — Detection must be instant and offline
• ❌ Recommending for non-existent files — Only recommend if file is confirmed present
• ❌ Over-recommending — Suggest 2-4 core skills per environment
• ❌ Ignoring skill composition — Include clean-code in every recommendation

KB Reference

~/vaults/baphled/3. Resources/Knowledge Base/AI Development System/Skills/Agent-Guidance/Core Auto Detect.md

Related skills

• clean-code — Applies across all detected environments
• automation — Complements build system detection
• devops — Complements CI/CD detection
• critical-thinking — For evaluating when to trust auto-detection vs manual selection