chore: update usage examples in workflow scripts

- Replace direct execution method with `uv run` in chore_implement.py
- Update prompt.py to use `uv run` for all usage examples
- Modify refactor.md to clarify refactoring instructions
- Add new state management functionality in state.py
- Introduce build and plan workflows with state management

Author: MarkusNeusinger

.claude/settings.json

@@ -14,5 +14,8 @@
       "Bash(git push *)"
     ],
     "deny": []
+  },
+  "env": {
+    "CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS": "1"
   }
 }

agentic/commands/audit.md

@@ -0,0 +1,217 @@
+# Code Quality Refactor
+
+> Comprehensive code quality analysis for the pyplots repository. Checks readability, performance, code smells, consistency, and modernization opportunities across all components.
+
+## Context
+
+@CLAUDE.md
+@pyproject.toml
+
+## Analysis Scope
+
+Analyze the following areas systematically. For each area, identify:
+- **Code Smells**: Dead code, duplication, overly complex functions, god classes
+- **Readability**: Unclear naming, missing/outdated comments, inconsistent formatting
+- **Performance**: Inefficient patterns, N+1 queries, unnecessary computations
+- **Modernization**: Deprecated APIs, old Python patterns, outdated dependencies
+- **Consistency**: Naming conventions, import styles, error handling patterns
+- **Type Safety**: Missing type hints, incorrect types, Any overuse
+- **Test Quality**: Missing tests, flaky tests, poor assertions
+
+## Discover Structure
+
+```bash
+# Overview of repository structure
+find . -type f -name "*.py" | grep -E "^\./(api|core|automation|tests)/" | head -80
+
+# Frontend structure
+find ./app/src -type f -name "*.ts" -o -name "*.tsx" 2>/dev/null | head -40
+
+# Workflow files
+ls -la .github/workflows/*.yml 2>/dev/null
+
+# Prompt files
+find ./prompts -type f -name "*.md" 2>/dev/null
+
+# Documentation
+find ./docs -type f -name "*.md" 2>/dev/null
+```
+
+## Run Quality Checks
+
+```bash
+# Linting issues
+uv run ruff check .
+
+# Format check (dry-run)
+uv run ruff format . --check --diff
+```
+
+## Analyze Backend (api/)
+
+Check for:
+- Router organization and REST conventions
+- Dependency injection patterns
+- Error handling consistency
+- Response schema completeness
+- Async/await correctness
+- Caching strategies
+
+```bash
+ls -la api/*.py api/**/*.py 2>/dev/null
+```
+
+## Analyze Core Logic (core/)
+
+Check for:
+- Repository pattern implementation
+- Database model design
+- Configuration management
+- Utility function organization
+- Type definitions
+
+```bash
+ls -la core/*.py core/**/*.py 2>/dev/null
+```
+
+## Analyze Automation (automation/)
+
+Check for:
+- Script modularity
+- Error handling in workflows
+- CLI argument parsing
+- Logging consistency
+
+```bash
+ls -la automation/**/*.py 2>/dev/null
+```
+
+## Analyze Frontend (app/src/)
+
+Check for:
+- Component structure and reusability
+- Hook patterns and custom hooks
+- TypeScript strictness (no `any`, proper interfaces)
+- Performance (memo, useCallback only where truly needed)
+- Accessibility (aria-labels, keyboard navigation, focus management)
+- Consistent error handling and error boundaries
+- Loading states and skeletons
+- Unused imports and dead code
+
+```bash
+ls -la app/src/**/*.tsx app/src/**/*.ts 2>/dev/null | head -30
+```
+
+## Analyze Tests (tests/)
+
+Check for:
+- Test coverage gaps
+- Fixture organization
+- Mock patterns
+- Assertion quality
+- Test naming conventions
+
+```bash
+find tests -name "*.py" -type f | head -30
+```
+
+## Analyze GitHub Workflows (.github/workflows/)
+
+Check for:
+- Workflow consistency and naming
+- Job dependencies and parallelization
+- Secret handling and security
+- Error handling and failure modes
+- Concurrency settings
+- Reusable workflows vs duplication
+- Environment variable consistency
+- Trigger conditions (labels, paths, branches)
+
+```bash
+ls -la .github/workflows/*.yml
+```
+
+## Analyze AI Prompts (prompts/)
+
+Check for:
+- Prompt clarity and structure
+- Consistency across prompt files
+- Outdated instructions or references
+- Missing edge cases
+- Template completeness
+- Library-specific rules alignment
+
+```bash
+find prompts -name "*.md" -o -name "*.yaml" | sort
+```
+
+## Analyze Documentation (docs/)
+
+Check for:
+- Outdated information vs actual code
+- Missing documentation for new features
+- Broken internal links
+- Inconsistent formatting
+- Sync with CLAUDE.md
+
+```bash
+find docs -name "*.md" | sort
+```
+
+## Cross-Check: Consistency
+
+Verify alignment across components:
+- Do workflows reference existing prompt files?
+- Are library names consistent across `prompts/library/`, `core/constants.py`, workflows?
+- Do workflow labels match documentation?
+- Are schema definitions in sync between API and frontend types?
+
+```bash
+# Check prompt references in workflows
+grep -r "prompts/" .github/workflows/ 2>/dev/null | head -20
+
+# Library definitions
+ls prompts/library/ 2>/dev/null | sed 's/.md//' | sort
+grep -E "SUPPORTED_LIBRARIES|LIBRARIES" core/constants.py 2>/dev/null | head -5
+```
+
+## Run Test Suite
+
+```bash
+# Quick test run to check for failures
+uv run pytest tests/unit -x -q --tb=no 2>/dev/null | tail -20
+```
+
+## Output Format
+
+After analysis, provide a structured report:
+
+### 1. Critical Issues (Must Fix)
+Issues that could cause bugs, security problems, or break functionality.
+
+### 2. High Priority (Should Fix)
+Code smells, performance issues, or maintainability concerns.
+
+### 3. Medium Priority (Consider)
+Modernization opportunities, style improvements, minor inconsistencies.
+
+### 4. Low Priority (Nice to Have)
+Minor improvements, documentation updates, cosmetic changes.
+
+### 5. Positive Patterns
+Good practices found that should be maintained or expanded.
+
+For each issue, provide:
+- **Location**: File path and line number
+- **Problem**: Clear description of the issue
+- **Impact**: Why this matters
+- **Solution**: Specific fix recommendation
+- **Effort**: Low/Medium/High
+
+## Exclusions
+
+Do NOT flag:
+- Plot implementations in `plots/` (AI-generated, different style rules)
+- Generated files or lock files (`uv.lock`, `yarn.lock`, etc.)
+- Third-party code
+- Issues already tracked in pyproject.toml exclusions

agentic/commands/classify.md

@@ -0,0 +1,22 @@
+# Task Classification
+
+Classify the task description into exactly one category.
+
+## Task
+
+$ARGUMENTS
+
+## Rules
+
+Analyze the task and classify it:
+
+- `bug` — Something is broken, errors occur, regressions, 404s, crashes, incorrect behavior, things that used to work
+- `feature` — New functionality, additions, enhancements, new pages, new endpoints, new capabilities
+- `chore` — Maintenance, cleanup, config changes, dependency updates, documentation, CI/CD, formatting
+- `refactor` — Restructuring code while preserving behavior: extracting modules, reorganizing files, splitting classes, renaming for clarity, moving code between files
+
+## Response
+
+Respond with ONLY a JSON object on a single line. No explanation, no markdown, no code fences:
+
+{"type": "bug", "reason": "one sentence explaining why"}