The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
docs: add repo evaluation vs current setup comparison
Claude4e7773c2ce18
diff --git a/EVALUATION.md b/EVALUATION.mdnew file mode 100644index 000000000..6ecc969c7--- /dev/null+++ b/EVALUATION.md@@ -0,0 +1,122 @@+# Repo Evaluation vs Current Setup++**Date:** 2026-03-21+**Branch:** `claude/evaluate-repo-comparison-ASZ9Y`++---++## Current Setup (`~/.claude/`)++The active Claude Code installation is near-minimal:++| Component | Current |+|-----------|---------|+| Agents | 0 |+| Skills | 0 installed |+| Commands | 0 |+| Hooks | 1 (Stop: git check) |+| Rules | 0 |+| MCP configs | 0 |++**Installed hooks:**+- `Stop` → `stop-hook-git-check.sh` — blocks session end if there are uncommitted changes or unpushed commits++**Installed permissions:**+- `Skill` — allows skill invocations++**Plugins:** Only `blocklist.json` (no active plugins installed)++---++## This Repo (`everything-claude-code` v1.9.0)++| Component | Repo |+|-----------|------|+| Agents | 28 |+| Skills | 116 |+| Commands | 59 |+| Rules sets | 12 languages + common (60+ rule files) |+| Hooks | Comprehensive system (PreToolUse, PostToolUse, SessionStart, Stop) |+| MCP configs | 1 (Context7 + others) |+| Schemas | 9 JSON validators |+| Scripts/CLI | 46+ Node.js modules + multiple CLIs |+| Tests | 58 test files |+| Install profiles | minimal, standard, full, enterprise |+| Supported harnesses | Claude Code, Codex, Cursor, OpenCode |++---++## Gap Analysis++### Hooks+- **Current:** 1 Stop hook (git hygiene check)+- **Repo:** Full hook matrix covering:+ - Dangerous command blocking (`rm -rf`, force pushes)+ - Auto-formatting on file edits+ - Dev server tmux enforcement+ - Cost tracking+ - Session evaluation and governance capture+ - MCP health monitoring++### Agents (28 missing)+The repo provides specialized agents for every major workflow:+- Language reviewers: TypeScript, Python, Go, Java, Kotlin, Rust, C++, Flutter+- Build resolvers: Go, Java, Kotlin, Rust, C++, PyTorch+- Workflow agents: planner, tdd-guide, code-reviewer, security-reviewer, architect+- Automation: loop-operator, doc-updater, refactor-cleaner, harness-optimizer++### Skills (116 missing)+Domain knowledge modules covering:+- Language patterns (Python, Go, Kotlin, Rust, C++, Java, Swift, Perl, Laravel, Django)+- Testing strategies (TDD, E2E, coverage)+- Architecture patterns (backend, frontend, API design, database migrations)+- AI/ML workflows (Claude API, eval harness, agent loops, cost-aware pipelines)+- Business workflows (investor materials, market research, content engine)++### Commands (59 missing)+- `/tdd`, `/plan`, `/e2e`, `/code-review` — core dev workflows+- `/sessions`, `/save-session`, `/resume-session` — session persistence+- `/orchestrate`, `/multi-plan`, `/multi-execute` — multi-agent coordination+- `/learn`, `/skill-create`, `/evolve` — continuous improvement+- `/build-fix`, `/verify`, `/quality-gate` — build/quality automation++### Rules (60+ files missing)+Language-specific coding style, patterns, testing, and security guidelines for:+TypeScript, Python, Go, Java, Kotlin, Rust, C++, C#, Swift, Perl, PHP, and common/cross-language rules.++---++## Recommendations++### Immediate value (minimal install)+Run `ecc install --profile minimal` to get:+- Core agents (code-reviewer, planner, tdd-guide, security-reviewer)+- Essential skills (tdd-workflow, coding-standards, security-review)+- Key commands (/tdd, /plan, /code-review, /build-fix)++### Full install+Run `ecc install --profile full` to get all 28 agents, 116 skills, and 59 commands.++### Hooks upgrade+The current Stop hook is solid. The repo's `hooks.json` adds:+- Dangerous command blocking (safety)+- Auto-formatting (quality)+- Cost tracking (observability)+- Session evaluation (learning)++### Rules+Adding language rules (e.g., TypeScript, Python) provides always-on coding guidelines without relying on per-session prompts.++---++## What the Current Setup Does Well++- The `stop-hook-git-check.sh` Stop hook is production-quality and already enforces good git hygiene+- The `Skill` permission is correctly configured+- The setup is clean with no conflicts or cruft++---++## Summary++The current setup is essentially a blank slate with one well-implemented git hygiene hook. This repo provides a complete, production-tested enhancement layer covering agents, skills, commands, hooks, and rules — with a selective install system so you can add exactly what you need without bloating the configuration.