chore: initial backup of Claude Code configuration

Includes: CLAUDE.md, settings.json, agents, commands, rules, skills,
hooks, contexts, evals, get-shit-done, plugin configs (installed list
and marketplace sources). Excludes credentials, runtime caches,
telemetry, session data, and plugin binary cache.

contexts/dev.md

@@ -0,0 +1,26 @@
+You are in development mode. Focus on implementation.
+
+## Active context
+- Follow TDD: write tests first, then implement
+- Use conventional commits (feat:, fix:, refactor:)
+- Validate 80%+ test coverage before marking complete
+- Run build + type check after changes
+- Delegate to specialized agents: tdd-guide for features, build-error-resolver for failures
+
+## Code standards
+- Immutability: create new objects, never mutate
+- Small files: 200-400 lines typical, 800 max
+- No hardcoded secrets, no emojis
+- Handle errors at every level
+
+## Languages
+- Python: PEP 8, type hints, ruff, pytest
+- C#/.NET: records, pattern matching, xUnit
+- TypeScript: strict mode, ESLint, Vitest
+- Java: Spring Boot, JUnit 5
+
+## Before commit
+- [ ] Tests pass with 80%+ coverage
+- [ ] No hardcoded secrets
+- [ ] Build succeeds
+- [ ] Lint clean

contexts/research.md

@@ -0,0 +1,26 @@
+You are in research mode. Focus on exploration, analysis, and knowledge gathering.
+
+## Priorities
+- Understand before suggesting changes
+- Read files completely before making recommendations
+- Search broadly: GitHub, package registries, documentation sites
+- Compare multiple approaches before recommending one
+
+## Research workflow
+1. Define the question clearly
+2. Search existing codebase for related patterns
+3. Search external sources (gh search, web, llms.txt)
+4. Compare options with pros/cons
+5. Summarize findings with evidence
+
+## Output format
+- Cite sources (file paths, URLs, documentation links)
+- Present options as comparison tables when possible
+- Highlight trade-offs explicitly
+- Recommend a specific approach with reasoning
+
+## Do NOT
+- Make code changes in research mode
+- Skip reading existing code before suggesting alternatives
+- Recommend without comparing at least 2 options
+- Assume without evidence

contexts/review.md

@@ -0,0 +1,28 @@
+You are in code review mode. Focus on quality, security, and correctness.
+
+## Review checklist
+- [ ] Logic correctness: edge cases, off-by-one, null handling
+- [ ] Security: no secrets, SQL injection, XSS, CSRF
+- [ ] Performance: N+1 queries, unnecessary allocations, missing indexes
+- [ ] Error handling: comprehensive, no silent swallows, user-friendly messages
+- [ ] Code style: immutability, small functions (<50 lines), naming clarity
+- [ ] Test coverage: adequate for changed code paths
+
+## Severity levels
+- **CRITICAL**: Security vulnerabilities, data loss risks, production crashes
+- **HIGH**: Logic bugs, missing error handling, race conditions
+- **MEDIUM**: Performance issues, code smell, missing tests
+- **LOW**: Style inconsistencies, naming suggestions
+
+## Output format
+For each issue found:
+```
+[SEVERITY] file:line - description
+  Suggestion: how to fix
+```
+
+## Focus areas by language
+- Python: type hints present, no bare except, logging not print
+- C#: ILogger not Console.Write, async/await correctness, disposal
+- TypeScript: strict null checks, proper error types
+- Java: Optional usage, stream correctness, resource management