kodelyth-ecc 1.2.0

package/skills/gan-style-harness/SKILL.md

@@ -0,0 +1,278 @@
+---
+name: gan-style-harness
+description: "GAN-inspired Generator-Evaluator agent harness for building high-quality applications autonomously. Based on Anthropic's March 2026 harness design paper."
+origin: ECC-community
+tools: Read, Write, Edit, Bash, Grep, Glob, Task
+---
+
+# GAN-Style Harness Skill
+
+> Inspired by [Anthropic's Harness Design for Long-Running Application Development](https://www.anthropic.com/engineering/harness-design-long-running-apps) (March 24, 2026)
+
+A multi-agent harness that separates **generation** from **evaluation**, creating an adversarial feedback loop that drives quality far beyond what a single agent can achieve.
+
+## Core Insight
+
+> When asked to evaluate their own work, agents are pathological optimists — they praise mediocre output and talk themselves out of legitimate issues. But engineering a **separate evaluator** to be ruthlessly strict is far more tractable than teaching a generator to self-critique.
+
+This is the same dynamic as GANs (Generative Adversarial Networks): the Generator produces, the Evaluator critiques, and that feedback drives the next iteration.
+
+## When to Use
+
+- Building complete applications from a one-line prompt
+- Frontend design tasks requiring high visual quality
+- Full-stack projects that need working features, not just code
+- Any task where "AI slop" aesthetics are unacceptable
+- Projects where you want to invest $50-200 for production-quality output
+
+## When NOT to Use
+
+- Quick single-file fixes (use standard `claude -p`)
+- Tasks with tight budget constraints (<$10)
+- Simple refactoring (use de-sloppify pattern instead)
+- Tasks that are already well-specified with tests (use TDD workflow)
+
+## Architecture
+
+```
+                    ┌─────────────┐
+                    │   PLANNER   │
+                    │  (Opus 4.6) │
+                    └──────┬──────┘
+                           │ Product Spec
+                           │ (features, sprints, design direction)
+                           ▼
+              ┌────────────────────────┐
+              │                        │
+              │   GENERATOR-EVALUATOR  │
+              │      FEEDBACK LOOP     │
+              │                        │
+              │  ┌──────────┐          │
+              │  │GENERATOR │--build-->│──┐
+              │  │(Opus 4.6)│          │  │
+              │  └────▲─────┘          │  │
+              │       │                │  │ live app
+              │    feedback             │  │
+              │       │                │  │
+              │  ┌────┴─────┐          │  │
+              │  │EVALUATOR │<-test----│──┘
+              │  │(Opus 4.6)│          │
+              │  │+Playwright│         │
+              │  └──────────┘          │
+              │                        │
+              │   5-15 iterations      │
+              └────────────────────────┘
+```
+
+## The Three Agents
+
+### 1. Planner Agent
+
+**Role:** Product manager — expands a brief prompt into a full product specification.
+
+**Key behaviors:**
+- Takes a one-line prompt and produces a 16-feature, multi-sprint specification
+- Defines user stories, technical requirements, and visual design direction
+- Is deliberately **ambitious** — conservative planning leads to underwhelming results
+- Produces evaluation criteria that the Evaluator will use later
+
+**Model:** Opus 4.6 (needs deep reasoning for spec expansion)
+
+### 2. Generator Agent
+
+**Role:** Developer — implements features according to the spec.
+
+**Key behaviors:**
+- Works in structured sprints (or continuous mode with newer models)
+- Negotiates a "sprint contract" with the Evaluator before writing code
+- Uses full-stack tooling: React, FastAPI/Express, databases, CSS
+- Manages git for version control between iterations
+- Reads Evaluator feedback and incorporates it in next iteration
+
+**Model:** Opus 4.6 (needs strong coding capability)
+
+### 3. Evaluator Agent
+
+**Role:** QA engineer — tests the live running application, not just code.
+
+**Key behaviors:**
+- Uses **Playwright MCP** to interact with the live application
+- Clicks through features, fills forms, tests API endpoints
+- Scores against four criteria (configurable):
+  1. **Design Quality** — Does it feel like a coherent whole?
+  2. **Originality** — Custom decisions vs. template/AI patterns?
+  3. **Craft** — Typography, spacing, animations, micro-interactions?
+  4. **Functionality** — Do all features actually work?
+- Returns structured feedback with scores and specific issues
+- Is engineered to be **ruthlessly strict** — never praises mediocre work
+
+**Model:** Opus 4.6 (needs strong judgment + tool use)
+
+## Evaluation Criteria
+
+The default four criteria, each scored 1-10:
+
+```markdown
+## Evaluation Rubric
+
+### Design Quality (weight: 0.3)
+- 1-3: Generic, template-like, "AI slop" aesthetics
+- 4-6: Competent but unremarkable, follows conventions
+- 7-8: Distinctive, cohesive visual identity
+- 9-10: Could pass for a professional designer's work
+
+### Originality (weight: 0.2)
+- 1-3: Default colors, stock layouts, no personality
+- 4-6: Some custom choices, mostly standard patterns
+- 7-8: Clear creative vision, unique approach
+- 9-10: Surprising, delightful, genuinely novel
+
+### Craft (weight: 0.3)
+- 1-3: Broken layouts, missing states, no animations
+- 4-6: Works but feels rough, inconsistent spacing
+- 7-8: Polished, smooth transitions, responsive
+- 9-10: Pixel-perfect, delightful micro-interactions
+
+### Functionality (weight: 0.2)
+- 1-3: Core features broken or missing
+- 4-6: Happy path works, edge cases fail
+- 7-8: All features work, good error handling
+- 9-10: Bulletproof, handles every edge case
+```
+
+### Scoring
+
+- **Weighted score** = sum of (criterion_score * weight)
+- **Pass threshold** = 7.0 (configurable)
+- **Max iterations** = 15 (configurable, typically 5-15 sufficient)
+
+## Usage
+
+### Via Command
+
+```bash
+# Full three-agent harness
+/project:gan-build "Build a project management app with Kanban boards, team collaboration, and dark mode"
+
+# With custom config
+/project:gan-build "Build a recipe sharing platform" --max-iterations 10 --pass-threshold 7.5
+
+# Frontend design mode (generator + evaluator only, no planner)
+/project:gan-design "Create a landing page for a crypto portfolio tracker"
+```
+
+### Via Shell Script
+
+```bash
+# Basic usage
+./scripts/gan-harness.sh "Build a music streaming dashboard"
+
+# With options
+GAN_MAX_ITERATIONS=10 \
+GAN_PASS_THRESHOLD=7.5 \
+GAN_EVAL_CRITERIA="functionality,performance,security" \
+./scripts/gan-harness.sh "Build a REST API for task management"
+```
+
+### Via Claude Code (Manual)
+
+```bash
+# Step 1: Plan
+claude -p --model opus "You are a Product Planner. Read PLANNER_PROMPT.md. Expand this brief into a full product spec: 'Build a Kanban board app'. Write spec to spec.md"
+
+# Step 2: Generate (iteration 1)
+claude -p --model opus "You are a Generator. Read spec.md. Implement Sprint 1. Start the dev server on port 3000."
+
+# Step 3: Evaluate (iteration 1)
+claude -p --model opus --allowedTools "Read,Bash,mcp__playwright__*" "You are an Evaluator. Read EVALUATOR_PROMPT.md. Test the live app at http://localhost:3000. Score against the rubric. Write feedback to feedback-001.md"
+
+# Step 4: Generate (iteration 2 — reads feedback)
+claude -p --model opus "You are a Generator. Read spec.md and feedback-001.md. Address all issues. Improve the scores."
+
+# Repeat steps 3-4 until pass threshold met
+```
+
+## Evolution Across Model Capabilities
+
+The harness should simplify as models improve. Following Anthropic's evolution:
+
+### Stage 1 — Weaker Models (Sonnet-class)
+- Full sprint decomposition required
+- Context resets between sprints (avoid context anxiety)
+- 2-agent minimum: Initializer + Coding Agent
+- Heavy scaffolding compensates for model limitations
+
+### Stage 2 — Capable Models (Opus 4.5-class)
+- Full 3-agent harness: Planner + Generator + Evaluator
+- Sprint contracts before each implementation phase
+- 10-sprint decomposition for complex apps
+- Context resets still useful but less critical
+
+### Stage 3 — Frontier Models (Opus 4.6-class)
+- Simplified harness: single planning pass, continuous generation
+- Evaluation reduced to single end-pass (model is smarter)
+- No sprint structure needed
+- Automatic compaction handles context growth
+
+> **Key principle:** Every harness component encodes an assumption about what the model can't do alone. When models improve, re-test those assumptions. Strip away what's no longer needed.
+
+## Configuration
+
+### Environment Variables
+
+| Variable | Default | Description |
+|----------|---------|-------------|
+| `GAN_MAX_ITERATIONS` | `15` | Maximum generator-evaluator cycles |
+| `GAN_PASS_THRESHOLD` | `7.0` | Weighted score to pass (1-10) |
+| `GAN_PLANNER_MODEL` | `opus` | Model for planning agent |
+| `GAN_GENERATOR_MODEL` | `opus` | Model for generator agent |
+| `GAN_EVALUATOR_MODEL` | `opus` | Model for evaluator agent |
+| `GAN_EVAL_CRITERIA` | `design,originality,craft,functionality` | Comma-separated criteria |
+| `GAN_DEV_SERVER_PORT` | `3000` | Port for the live app |
+| `GAN_DEV_SERVER_CMD` | `npm run dev` | Command to start dev server |
+| `GAN_PROJECT_DIR` | `.` | Project working directory |
+| `GAN_SKIP_PLANNER` | `false` | Skip planner, use spec directly |
+| `GAN_EVAL_MODE` | `playwright` | `playwright`, `screenshot`, or `code-only` |
+
+### Evaluation Modes
+
+| Mode | Tools | Best For |
+|------|-------|----------|
+| `playwright` | Browser MCP + live interaction | Full-stack apps with UI |
+| `screenshot` | Screenshot + visual analysis | Static sites, design-only |
+| `code-only` | Tests + linting + build | APIs, libraries, CLI tools |
+
+## Anti-Patterns
+
+1. **Evaluator too lenient** — If the evaluator passes everything on iteration 1, your rubric is too generous. Tighten scoring criteria and add explicit penalties for common AI patterns.
+
+2. **Generator ignoring feedback** — Ensure feedback is passed as a file, not inline. The generator should read `feedback-NNN.md` at the start of each iteration.
+
+3. **Infinite loops** — Always set `GAN_MAX_ITERATIONS`. If the generator can't improve past a score plateau after 3 iterations, stop and flag for human review.
+
+4. **Evaluator testing superficially** — The evaluator must use Playwright to **interact** with the live app, not just screenshot it. Click buttons, fill forms, test error states.
+
+5. **Evaluator praising its own fixes** — Never let the evaluator suggest fixes and then evaluate those fixes. The evaluator only critiques; the generator fixes.
+
+6. **Context exhaustion** — For long sessions, use Claude Agent SDK's automatic compaction or reset context between major phases.
+
+## Results: What to Expect
+
+Based on Anthropic's published results:
+
+| Metric | Solo Agent | GAN Harness | Improvement |
+|--------|-----------|-------------|-------------|
+| Time | 20 min | 4-6 hours | 12-18x longer |
+| Cost | $9 | $125-200 | 14-22x more |
+| Quality | Barely functional | Production-ready | Phase change |
+| Core features | Broken | All working | N/A |
+| Design | Generic AI slop | Distinctive, polished | N/A |
+
+**The tradeoff is clear:** ~20x more time and cost for a qualitative leap in output quality. This is for projects where quality matters.
+
+## References
+
+- [Anthropic: Harness Design for Long-Running Apps](https://www.anthropic.com/engineering/harness-design-long-running-apps) — Original paper by Prithvi Rajasekaran
+- [Epsilla: The GAN-Style Agent Loop](https://www.epsilla.com/blogs/anthropic-harness-engineering-multi-agent-gan-architecture) — Architecture deconstruction
+- [Martin Fowler: Harness Engineering](https://martinfowler.com/articles/exploring-gen-ai/harness-engineering.html) — Broader industry context
+- [OpenAI: Harness Engineering](https://openai.com/index/harness-engineering/) — OpenAI's parallel work

package/skills/git-mastery/SKILL.md

@@ -0,0 +1,366 @@
+---
+name: git-mastery
+description: Advanced Git workflows, branching strategies, history management, and team collaboration patterns. Covers trunk-based development, interactive rebase, bisect debugging, monorepo workflows, and professional commit hygiene. Powered by Kodelyth.
+origin: Kodelyth
+---
+
+# Git Mastery — Advanced Git Workflows
+
+Professional-grade Git patterns for individuals and teams. Powered by Kodelyth.
+
+## When to Use
+
+- Setting up a branching strategy for a team
+- Learning advanced Git commands (rebase, bisect, worktree)
+- Fixing a messy commit history before merging
+- Debugging when a bug was introduced using `git bisect`
+- Working in a monorepo
+- Enforcing commit message standards
+
+---
+
+## Branching Strategies
+
+### Trunk-Based Development (Recommended for most teams)
+
+```
+main (always deployable)
+ ├── feat/add-oauth      (short-lived, < 2 days)
+ ├── fix/null-check      (short-lived, < 1 day)
+ └── chore/upgrade-deps  (short-lived)
+```
+
+**Rules:**
+- All branches off `main`, all PRs back to `main`
+- Branches live < 2 days — longer = higher merge conflict risk
+- Use feature flags to ship incomplete features safely
+- Every commit to `main` is potentially deployable
+
+```bash
+# Start a feature
+git checkout main && git pull
+git checkout -b feat/user-authentication
+
+# Sync with main daily (rebase, don't merge)
+git fetch origin
+git rebase origin/main
+
+# PR → squash merge → delete branch
+```
+
+### Gitflow (For versioned software with release cycles)
+
+```
+main          (production releases only)
+develop       (integration branch)
+ ├── feature/xyz     (features off develop)
+ ├── release/1.2.0   (stabilization)
+ └── hotfix/critical (off main, merges to main + develop)
+```
+
+**Use when**: You ship versioned releases (mobile apps, libraries, enterprise software)
+**Avoid when**: You deploy continuously (web apps, SaaS)
+
+### Ship / Show / Ask (For experienced teams)
+
+```
+Ship: Commit directly to main (trivial changes, docs, chores)
+Show: Open PR, merge immediately without review (small, safe changes)
+Ask:  Open PR, wait for review (significant changes, risk)
+```
+
+---
+
+## Commit Message Mastery
+
+### Conventional Commits Format
+
+```
+<type>(<scope>): <description>
+
+[optional body]
+
+[optional footer]
+```
+
+**Types:**
+```
+feat:     New feature (triggers MINOR version bump)
+fix:      Bug fix (triggers PATCH version bump)
+docs:     Documentation only
+style:    Formatting (no logic change)
+refactor: Code restructure (no feature, no fix)
+test:     Adding or fixing tests
+chore:    Build, tooling, dependencies
+perf:     Performance improvement
+ci:       CI/CD changes
+revert:   Reverting a commit
+```
+
+**Examples:**
+```bash
+# Good
+git commit -m "feat(auth): add OAuth2 login with Google"
+git commit -m "fix(cart): prevent double-submission on slow networks"
+git commit -m "refactor(user): extract UserService from UserController"
+git commit -m "test(payment): add integration test for failed webhooks"
+
+# Bad
+git commit -m "fix bug"
+git commit -m "WIP"
+git commit -m "changes"
+git commit -m "asdf"
+```
+
+**With body (for significant changes):**
+```bash
+git commit -m "feat(billing): add Stripe subscription management
+
+Adds three subscription tiers: Free, Pro ($29/mo), Enterprise ($99/mo).
+Payment flow uses Stripe Checkout. Subscription status synced via webhooks.
+
+Breaking: removes legacy PayPal integration (deprecated since v1.3)
+Closes #234"
+```
+
+---
+
+## Interactive Rebase — History Surgery
+
+### Squash WIP commits before merging
+
+```bash
+# You have 5 messy commits you want to clean up
+git log --oneline -5
+# abc1234 fix
+# def5678 WIP
+# ghi9012 more changes
+# jkl3456 actually fix it
+# mno7890 feat: add user search
+
+# Squash all 5 into 1 clean commit
+git rebase -i HEAD~5
+
+# In the editor:
+# pick mno7890 feat: add user search
+# squash jkl3456 actually fix it
+# squash ghi9012 more changes
+# squash def5678 WIP
+# squash abc1234 fix
+# → writes a clean single commit message
+```
+
+### Fix a commit message (already pushed? see below)
+
+```bash
+# Fix the last commit message
+git commit --amend -m "feat(search): add full-text user search with pagination"
+
+# Fix a message 3 commits back
+git rebase -i HEAD~3
+# Change "pick" to "reword" on the target commit
+```
+
+### Fixup — add changes to a previous commit
+
+```bash
+# You forgot to add a file to a commit 2 commits ago
+git add forgotten-file.ts
+git commit --fixup HEAD~2
+
+# Then squash fixup into its target automatically
+git rebase -i --autosquash HEAD~3
+```
+
+### Split a large commit into smaller ones
+
+```bash
+git rebase -i HEAD~1
+# Change "pick" to "edit" on the commit to split
+
+# Now git stops at that commit
+git reset HEAD~1         # unstage all changes
+git add -p               # selectively stage first logical change
+git commit -m "feat: add user model"
+git add -p               # stage second logical change
+git commit -m "feat: add user repository"
+git rebase --continue    # finish the rebase
+```
+
+---
+
+## Git Bisect — Binary Search for Bugs
+
+When you know "it worked in v1.2, broke somewhere before v1.5":
+
+```bash
+git bisect start
+git bisect bad                    # current HEAD is broken
+git bisect good v1.2.0            # this tag/commit was good
+
+# Git checks out the midpoint automatically
+# Test it (run your test, check the behavior)
+git bisect good   # if this commit is OK
+git bisect bad    # if this commit has the bug
+
+# Repeat — git halves the search space each time
+# With 1000 commits, it takes at most 10 steps
+
+# Automate with a test script
+git bisect run npm test -- --grep "the failing test"
+
+# When done
+git bisect reset  # returns to original HEAD
+```
+
+---
+
+## Git Worktrees — Multiple Branches Simultaneously
+
+Work on a hotfix while keeping your feature branch untouched:
+
+```bash
+# Create a new worktree (separate directory, same repo)
+git worktree add ../hotfix-payment hotfix/payment-crash
+
+# Now you have two working directories:
+#   ~/project/           → your feature branch
+#   ~/hotfix-payment/    → the hotfix branch
+
+# Work in the hotfix directory, commit, push, PR
+cd ../hotfix-payment
+# ... make fix ...
+git commit -m "fix(payment): prevent double charge on retry"
+git push origin hotfix/payment-crash
+
+# Clean up when done
+cd ~/project
+git worktree remove ../hotfix-payment
+```
+
+---
+
+## Git Log — Find Anything in History
+
+```bash
+# See changes to a specific file over time
+git log --follow --oneline -- src/components/Auth.tsx
+
+# See who last changed each line (with commit + author)
+git blame src/components/Auth.tsx
+
+# Search commit messages
+git log --oneline --grep="payment"
+
+# Search code changes (find when a string was added or removed)
+git log -S "functionName" --oneline
+
+# See what changed between two tags
+git log v1.2.0..v1.3.0 --oneline
+
+# Visualize branch graph
+git log --oneline --graph --decorate --all
+```
+
+---
+
+## Git Stash — Save Work in Progress
+
+```bash
+# Stash everything (tracked + untracked)
+git stash push -u -m "WIP: halfway through auth refactor"
+
+# List stashes
+git stash list
+# stash@{0}: WIP: halfway through auth refactor
+# stash@{1}: quick fix attempt
+
+# Apply and remove the latest stash
+git stash pop
+
+# Apply a specific stash (keep it)
+git stash apply stash@{1}
+
+# Stash only specific files
+git stash push -m "partial" -- src/auth.ts src/user.ts
+
+# Drop a stash
+git stash drop stash@{1}
+```
+
+---
+
+## Monorepo Git Patterns
+
+```bash
+# Only run CI for changed packages
+git diff --name-only origin/main...HEAD | grep "^packages/"
+
+# Sparse checkout — only clone part of a large monorepo
+git clone --filter=blob:none --sparse https://github.com/org/monorepo
+git sparse-checkout init --cone
+git sparse-checkout set packages/my-package
+
+# Tag per package
+git tag packages/auth@1.2.0
+git tag packages/api@2.0.1
+
+# Find which package a file belongs to
+git log --oneline -- packages/billing/
+```
+
+---
+
+## Branch Protection Rules (GitHub/GitLab)
+
+```yaml
+# Recommended branch protection for main:
+required_status_checks:
+  strict: true           # must be up to date with main
+  contexts:
+    - "CI / test"
+    - "CI / lint"
+    - "CI / typecheck"
+
+required_pull_request_reviews:
+  required_approving_review_count: 1
+  dismiss_stale_reviews: true
+  require_code_owner_reviews: true
+
+restrictions:
+  push: []               # nobody can push directly to main
+  force_push: false      # never allow force push to main
+  delete: false          # never allow delete of main
+```
+
+---
+
+## Emergency Recovery
+
+```bash
+# Accidentally deleted a branch?
+git reflog                           # find the last commit hash
+git checkout -b recovered-branch <hash>
+
+# Accidentally force pushed?
+git reflog origin/main              # see the hash before the force push
+git push origin <hash>:main --force # restore it
+
+# Committed to the wrong branch?
+git log --oneline -3                # note the commit hash
+git reset HEAD~1 --soft             # undo commit, keep changes staged
+git stash                           # stash the changes
+git checkout correct-branch
+git stash pop                       # apply changes to correct branch
+git commit -m "feat: the actual message"
+
+# Accidentally committed secrets?
+# 1. Rotate the secret IMMEDIATELY (assume it's compromised)
+# 2. Remove from history:
+git filter-repo --path secrets.env --invert-paths
+# OR use BFG Repo Cleaner for simpler cases
+```
+
+---
+
+> Powered by Kodelyth — Git is not just version control, it's your project's memory.