compact-agent 1.24.2 → 1.25.0

package/dist/rules.js

@@ -1,6 +1,6 @@
 /**
  * Rules engine — per-language coding standards.
- * Loads rules from ~/.crowcoder/rules/ and injects into system prompt.
+ * Loads rules from ~/.compact-agent/rules/ and injects into system prompt.
  * Ships with built-in presets for common languages.
  */
 import { readFileSync, writeFileSync, mkdirSync, existsSync, readdirSync } from 'node:fs';
@@ -10,126 +10,126 @@ import { getConfigDir } from './config.js';
 const RULES_DIR = join(getConfigDir(), 'rules');
 // ── Built-in rule presets ─────────────────────────────────
 const BUILTIN_RULES = {
-    typescript: `# TypeScript Rules
-- Use strict mode; enable all strict compiler options
-- Prefer const over let; never use var
-- Use explicit return types on exported functions
-- Use interfaces for object shapes, types for unions/intersections
-- Prefer readonly arrays and properties where possible
-- Use nullish coalescing (??) over OR (||) for defaults
-- Use optional chaining (?.) for deep property access
-- Handle all Promise rejections — no floating promises
-- Use template literals over string concatenation
-- Use ESM imports (import/export), not CommonJS (require)
-- File naming: kebab-case for files, PascalCase for components/classes
-- Max file length: 300 lines — split if larger
+    typescript: `# TypeScript Rules
+- Use strict mode; enable all strict compiler options
+- Prefer const over let; never use var
+- Use explicit return types on exported functions
+- Use interfaces for object shapes, types for unions/intersections
+- Prefer readonly arrays and properties where possible
+- Use nullish coalescing (??) over OR (||) for defaults
+- Use optional chaining (?.) for deep property access
+- Handle all Promise rejections — no floating promises
+- Use template literals over string concatenation
+- Use ESM imports (import/export), not CommonJS (require)
+- File naming: kebab-case for files, PascalCase for components/classes
+- Max file length: 300 lines — split if larger
 - Prefer early returns over deeply nested conditionals`,
-    python: `# Python Rules
-- Follow PEP 8 style guidelines
-- Use type hints on all function signatures
-- Use dataclasses or Pydantic for structured data
-- Prefer f-strings over .format() or % formatting
-- Use pathlib.Path instead of os.path for file operations
-- Use context managers (with) for resource management
-- Use list/dict/set comprehensions where readable
-- Prefer enumerate() over range(len())
-- Never use mutable default arguments (use None + factory)
-- Use logging module, not print(), for non-user-facing output
-- Max function length: 30 lines — extract if larger
+    python: `# Python Rules
+- Follow PEP 8 style guidelines
+- Use type hints on all function signatures
+- Use dataclasses or Pydantic for structured data
+- Prefer f-strings over .format() or % formatting
+- Use pathlib.Path instead of os.path for file operations
+- Use context managers (with) for resource management
+- Use list/dict/set comprehensions where readable
+- Prefer enumerate() over range(len())
+- Never use mutable default arguments (use None + factory)
+- Use logging module, not print(), for non-user-facing output
+- Max function length: 30 lines — extract if larger
 - Write docstrings for all public functions (Google style)`,
-    go: `# Go Rules
-- Follow Effective Go and Go Proverbs
-- Use gofmt/goimports for formatting
-- Handle every error — no _ for errors unless explicitly justified
-- Use short variable declarations (:=) inside functions
-- Keep interfaces small — 1-3 methods, defined where used
-- Use context.Context as the first parameter for long-running ops
-- Use table-driven tests
-- Prefer returning errors over panicking
-- Use defer for cleanup
-- Package naming: short, lowercase, no underscores
-- Avoid init() functions — prefer explicit initialization
+    go: `# Go Rules
+- Follow Effective Go and Go Proverbs
+- Use gofmt/goimports for formatting
+- Handle every error — no _ for errors unless explicitly justified
+- Use short variable declarations (:=) inside functions
+- Keep interfaces small — 1-3 methods, defined where used
+- Use context.Context as the first parameter for long-running ops
+- Use table-driven tests
+- Prefer returning errors over panicking
+- Use defer for cleanup
+- Package naming: short, lowercase, no underscores
+- Avoid init() functions — prefer explicit initialization
 - Use struct embedding for composition, not inheritance`,
-    rust: `# Rust Rules
-- Follow Rust API Guidelines (RFC 430)
-- Use clippy lints: #![warn(clippy::all)]
-- Prefer &str over String for function parameters
-- Use Result<T, E> for fallible operations, not panics
-- Implement Display for error types
-- Use iterators and combinators over manual loops where clear
-- Prefer owned types in struct fields, borrows in function params
-- Use derive macros for Debug, Clone, PartialEq where appropriate
-- Keep unsafe blocks minimal and well-documented
-- Use cargo fmt for formatting
+    rust: `# Rust Rules
+- Follow Rust API Guidelines (RFC 430)
+- Use clippy lints: #![warn(clippy::all)]
+- Prefer &str over String for function parameters
+- Use Result<T, E> for fallible operations, not panics
+- Implement Display for error types
+- Use iterators and combinators over manual loops where clear
+- Prefer owned types in struct fields, borrows in function params
+- Use derive macros for Debug, Clone, PartialEq where appropriate
+- Keep unsafe blocks minimal and well-documented
+- Use cargo fmt for formatting
 - Prefer match over if-let chains for >2 variants`,
-    java: `# Java Rules
-- Follow Google Java Style Guide
-- Use records for value objects (Java 16+)
-- Use var for local variables with clear initialization
-- Prefer List.of(), Map.of() over mutable collections
-- Use Optional instead of null for potentially absent values
-- Use try-with-resources for all Closeable resources
-- Prefer streams over explicit loops for collection transforms
-- Use @Override annotation always
-- Final fields by default; minimize mutability
-- One class per file; class name matches filename
+    java: `# Java Rules
+- Follow Google Java Style Guide
+- Use records for value objects (Java 16+)
+- Use var for local variables with clear initialization
+- Prefer List.of(), Map.of() over mutable collections
+- Use Optional instead of null for potentially absent values
+- Use try-with-resources for all Closeable resources
+- Prefer streams over explicit loops for collection transforms
+- Use @Override annotation always
+- Final fields by default; minimize mutability
+- One class per file; class name matches filename
 - Use SLF4J for logging`,
-    kotlin: `# Kotlin Rules
-- Follow Kotlin Coding Conventions
-- Use data classes for value objects
-- Prefer val over var — immutability by default
-- Use sealed classes for restricted hierarchies
-- Use when instead of if-else chains (>2 branches)
-- Use scope functions (let, run, apply, also) appropriately
-- Use coroutines for async — avoid callbacks
-- Prefer extension functions for utility operations
-- Use string templates over concatenation
+    kotlin: `# Kotlin Rules
+- Follow Kotlin Coding Conventions
+- Use data classes for value objects
+- Prefer val over var — immutability by default
+- Use sealed classes for restricted hierarchies
+- Use when instead of if-else chains (>2 branches)
+- Use scope functions (let, run, apply, also) appropriately
+- Use coroutines for async — avoid callbacks
+- Prefer extension functions for utility operations
+- Use string templates over concatenation
 - Use named arguments for functions with >3 parameters`,
-    cpp: `# C++ Rules
-- Follow C++ Core Guidelines
-- Use smart pointers (unique_ptr, shared_ptr) — no raw owning pointers
-- Use RAII for resource management
-- Prefer references over pointers where nullability is not needed
-- Use const correctly and consistently
-- Use auto for complex types where the type is obvious from context
-- Prefer range-based for loops
-- Use std::string_view for non-owning string parameters
-- Use [[nodiscard]] for functions where ignoring return is an error
-- Use std::optional for values that may not exist
-- Keep headers minimal — forward-declare where possible
+    cpp: `# C++ Rules
+- Follow C++ Core Guidelines
+- Use smart pointers (unique_ptr, shared_ptr) — no raw owning pointers
+- Use RAII for resource management
+- Prefer references over pointers where nullability is not needed
+- Use const correctly and consistently
+- Use auto for complex types where the type is obvious from context
+- Prefer range-based for loops
+- Use std::string_view for non-owning string parameters
+- Use [[nodiscard]] for functions where ignoring return is an error
+- Use std::optional for values that may not exist
+- Keep headers minimal — forward-declare where possible
 - Use namespaces to avoid name collisions`,
-    php: `# PHP Rules
-- Use PHP 8.1+ features: enums, fibers, readonly properties
-- Use strict types: declare(strict_types=1) in every file
-- Use type declarations for all parameters, return types, and properties
-- Follow PSR-12 coding style
-- Use constructor promotion for simple classes
-- Use match() over switch for value mapping
-- Use null coalescing (??) and nullsafe (?->) operators
-- Use named arguments for clarity
-- Prefer arrays + array functions over manual loops
+    php: `# PHP Rules
+- Use PHP 8.1+ features: enums, fibers, readonly properties
+- Use strict types: declare(strict_types=1) in every file
+- Use type declarations for all parameters, return types, and properties
+- Follow PSR-12 coding style
+- Use constructor promotion for simple classes
+- Use match() over switch for value mapping
+- Use null coalescing (??) and nullsafe (?->) operators
+- Use named arguments for clarity
+- Prefer arrays + array functions over manual loops
 - Use Composer autoloading (PSR-4)`,
-    sql: `# SQL Rules
-- Use parameterized queries to prevent injection
-- Avoid SELECT * — specify needed columns
-- Use proper indexes on WHERE, JOIN, ORDER BY columns
-- Avoid N+1 query patterns — use joins or eager loading
-- Write migrations that are reversible
-- Use foreign keys for referential integrity
-- Follow naming conventions: snake_case for tables/columns
-- Add CHECK constraints for business rule validation
-- Use transactions for multi-step operations
+    sql: `# SQL Rules
+- Use parameterized queries to prevent injection
+- Avoid SELECT * — specify needed columns
+- Use proper indexes on WHERE, JOIN, ORDER BY columns
+- Avoid N+1 query patterns — use joins or eager loading
+- Write migrations that are reversible
+- Use foreign keys for referential integrity
+- Follow naming conventions: snake_case for tables/columns
+- Add CHECK constraints for business rule validation
+- Use transactions for multi-step operations
 - Document complex queries with comments`,
-    csharp: `# C# Rules
-- Follow Microsoft C# Coding Conventions
-- Use PascalCase for public members, camelCase for private
-- Use LINQ where appropriate over manual loops
-- Use async/await for I/O operations
-- Use nullable reference types (C# 8+)
-- Prefer pattern matching in switch statements
-- Use dependency injection for testability
-- Follow SOLID principles
-- Use records for immutable data types
+    csharp: `# C# Rules
+- Follow Microsoft C# Coding Conventions
+- Use PascalCase for public members, camelCase for private
+- Use LINQ where appropriate over manual loops
+- Use async/await for I/O operations
+- Use nullable reference types (C# 8+)
+- Prefer pattern matching in switch statements
+- Use dependency injection for testability
+- Follow SOLID principles
+- Use records for immutable data types
 - Document public APIs with XML comments`,
 };
 function ensureDir() {
@@ -330,697 +330,697 @@ export function getLanguageRules(language) {
     }
     // Return comprehensive built-in rules with additional detail
     const detailedRules = {
-        typescript: `# TypeScript Code Standards
-
-## Type System & Strict Mode
-- Enforce \`"strict": true\` in tsconfig.json for all strict compiler options
-- Prefer const over let; never use var for any reason
-- Explicitly declare return types on all exported functions and public methods
-- Use interfaces for object shapes and contracts; use types for unions, intersections, and aliases
-- Prefer readonly arrays and properties to prevent accidental mutations
-- Always provide explicit generic type parameters; don't rely on inference for complex types
-- Use discriminated unions for type-safe pattern matching
-
-## Null & Undefined Safety
-- Use nullish coalescing (??) over logical OR (||) for proper falsy value handling
-- Use optional chaining (?.) for deep property access
-- Avoid non-null assertions (!) unless absolutely necessary and well-documented
-- Handle all Promise rejections—no floating promises
-- Never assume values exist; always add type guards or assertions
-
-## Async Patterns & Promises
-- Mark async operations with explicit async/await or return Promise<T>
-- Handle all .catch() blocks; use try-catch in async functions
-- Ensure Promise.all() and Promise.race() have error handling
-- Set timeouts on long-running async operations
-- Avoid mixing callbacks with promises
-
-## Imports & Module System
-- Use ESM (import/export); never CommonJS (require)
-- Import order: Node.js built-ins → npm packages → local files
-- Use named imports when importing specific exports
-- Avoid circular dependencies—refactor shared logic to a common module
-- Include file extensions in relative imports (.js, .ts)
-- Use index.ts for directory exports to control public API
-
-## Code Organization & Naming
-- Max function length: 50 lines; extract if larger
-- Max file length: 300 lines; split into multiple files if exceeded
-- Use kebab-case for filenames, PascalCase for classes/components
-- Use camelCase for functions and variables
-- Use UPPER_SNAKE_CASE for constants
-- Place public functions before private in file order
-
-## Code Quality & Maintainability
-- No console.log in production—use a logger (winston, pino, etc.)
-- Avoid deeply nested conditionals; use early returns for guard clauses
-- DRY: extract duplicated logic into reusable functions
-- No magic numbers—use named constants with descriptive names
-- Add comments explaining "why", not "what"
-- Keep cyclomatic complexity reasonable (<5 per function)
-
-## Performance & Memory
-- Avoid N+1 database queries by batch loading or joining data
-- Memoize expensive computations (use memoization libraries)
-- Remove event listeners (removeEventListener) to prevent memory leaks
-- Unsubscribe from subscriptions in cleanup functions
-- Don't hold large objects in memory longer than necessary
-
-## Testing
-- Write unit tests for all business logic and utilities
-- Cover happy paths and at least 3 edge cases per function
-- Add JSDoc comments to public functions with @param, @returns, @throws
+        typescript: `# TypeScript Code Standards
+
+## Type System & Strict Mode
+- Enforce \`"strict": true\` in tsconfig.json for all strict compiler options
+- Prefer const over let; never use var for any reason
+- Explicitly declare return types on all exported functions and public methods
+- Use interfaces for object shapes and contracts; use types for unions, intersections, and aliases
+- Prefer readonly arrays and properties to prevent accidental mutations
+- Always provide explicit generic type parameters; don't rely on inference for complex types
+- Use discriminated unions for type-safe pattern matching
+
+## Null & Undefined Safety
+- Use nullish coalescing (??) over logical OR (||) for proper falsy value handling
+- Use optional chaining (?.) for deep property access
+- Avoid non-null assertions (!) unless absolutely necessary and well-documented
+- Handle all Promise rejections—no floating promises
+- Never assume values exist; always add type guards or assertions
+
+## Async Patterns & Promises
+- Mark async operations with explicit async/await or return Promise<T>
+- Handle all .catch() blocks; use try-catch in async functions
+- Ensure Promise.all() and Promise.race() have error handling
+- Set timeouts on long-running async operations
+- Avoid mixing callbacks with promises
+
+## Imports & Module System
+- Use ESM (import/export); never CommonJS (require)
+- Import order: Node.js built-ins → npm packages → local files
+- Use named imports when importing specific exports
+- Avoid circular dependencies—refactor shared logic to a common module
+- Include file extensions in relative imports (.js, .ts)
+- Use index.ts for directory exports to control public API
+
+## Code Organization & Naming
+- Max function length: 50 lines; extract if larger
+- Max file length: 300 lines; split into multiple files if exceeded
+- Use kebab-case for filenames, PascalCase for classes/components
+- Use camelCase for functions and variables
+- Use UPPER_SNAKE_CASE for constants
+- Place public functions before private in file order
+
+## Code Quality & Maintainability
+- No console.log in production—use a logger (winston, pino, etc.)
+- Avoid deeply nested conditionals; use early returns for guard clauses
+- DRY: extract duplicated logic into reusable functions
+- No magic numbers—use named constants with descriptive names
+- Add comments explaining "why", not "what"
+- Keep cyclomatic complexity reasonable (<5 per function)
+
+## Performance & Memory
+- Avoid N+1 database queries by batch loading or joining data
+- Memoize expensive computations (use memoization libraries)
+- Remove event listeners (removeEventListener) to prevent memory leaks
+- Unsubscribe from subscriptions in cleanup functions
+- Don't hold large objects in memory longer than necessary
+
+## Testing
+- Write unit tests for all business logic and utilities
+- Cover happy paths and at least 3 edge cases per function
+- Add JSDoc comments to public functions with @param, @returns, @throws
 - Document non-obvious logic with inline comments`,
-        python: `# Python Code Standards
-
-## Type Hints & Static Typing
-- Add type hints to all function parameters and return types
-- Use type hints on class attributes using typing module or PEP 526
-- Use complex type aliases (TypeAlias) for clarity
-- Run mypy or pyright to validate types
-- Use Union[X, Y] or X | Y (3.10+) for multiple types
-- Never use implicit Any—always be explicit
-
-## PEP 8 Compliance
-- Use 4 spaces for indentation; never tabs
-- Keep lines ≤79 characters (100 for comments/docstrings)
-- Use CamelCase for classes; snake_case for functions and variables
-- Use UPPER_SNAKE_CASE for module-level constants
-- Place 2 blank lines between top-level definitions, 1 between methods
-- Use trailing commas in multi-line collections
-
-## Strings & Documentation
-- Always use f-strings (f"...") for string formatting
-- Never use % formatting or .format() unless f-string unavailable
-- Write docstrings in Google style for all public functions/classes
-- Use triple double quotes (""") for docstrings
-- Use raw strings (r"...") for regex patterns
-- Avoid string concatenation in loops—collect in list, then join
-
-## Async & Concurrency
-- Use async/await; don't mix with threading without careful consideration
-- Manage asyncio event loop explicitly; understand run_until_complete
-- Never block I/O in async functions (no time.sleep(); use asyncio.sleep())
-- Set timeouts on all async operations using asyncio.wait_for()
-- Use asyncio.gather() or asyncio.TaskGroup for concurrent tasks
-- Understand GIL implications when using threading
-
-## Exception Handling
-- Never use bare except:; always catch specific exceptions
-- Use custom exception classes for domain errors
-- Use finally blocks for cleanup or context managers (with statement)
-- Log exceptions with context before re-raising
-- Don't silently swallow exceptions
-- Use context managers for all resource management
-
-## File I/O & Paths
-- Always use pathlib.Path instead of os.path
-- Use context managers for file operations: with open(...) as f:
-- Always specify encoding explicitly (utf-8 is typical)
-- Never hardcode absolute paths—use config or environment variables
-- Check file existence before operations using if path.exists()
-
-## Data Structures & Functions
-- Limit function length to ~30 lines; extract if larger
-- Limit function parameters to 3-4; use dataclass if more needed
-- Never use mutable defaults: def foo(items=[]) is wrong; use None + factory
-- Prefer dataclasses or Pydantic for structured data
-- Use enumerate() instead of range(len())
-- Use list/dict/set comprehensions for readability
-- Use generator expressions for large datasets
-
-## Testing & Documentation
-- Write docstrings for all public functions using Google style
-- Use pytest for unit tests; organize with fixtures and markers
-- Cover happy paths, edge cases, and error conditions
-- Use mocks/patches to isolate external dependencies
-- Never use print() in production—use logging module
-- Configure logging with proper levels: DEBUG, INFO, WARNING, ERROR
-
-## Performance & Best Practices
-- Avoid N+1 queries using select_related() or prefetch_related() in ORMs
-- Use pagination for large result sets
-- Cache expensive computations (consider functools.lru_cache)
-- Avoid algorithms with O(n²) or worse without justification
+        python: `# Python Code Standards
+
+## Type Hints & Static Typing
+- Add type hints to all function parameters and return types
+- Use type hints on class attributes using typing module or PEP 526
+- Use complex type aliases (TypeAlias) for clarity
+- Run mypy or pyright to validate types
+- Use Union[X, Y] or X | Y (3.10+) for multiple types
+- Never use implicit Any—always be explicit
+
+## PEP 8 Compliance
+- Use 4 spaces for indentation; never tabs
+- Keep lines ≤79 characters (100 for comments/docstrings)
+- Use CamelCase for classes; snake_case for functions and variables
+- Use UPPER_SNAKE_CASE for module-level constants
+- Place 2 blank lines between top-level definitions, 1 between methods
+- Use trailing commas in multi-line collections
+
+## Strings & Documentation
+- Always use f-strings (f"...") for string formatting
+- Never use % formatting or .format() unless f-string unavailable
+- Write docstrings in Google style for all public functions/classes
+- Use triple double quotes (""") for docstrings
+- Use raw strings (r"...") for regex patterns
+- Avoid string concatenation in loops—collect in list, then join
+
+## Async & Concurrency
+- Use async/await; don't mix with threading without careful consideration
+- Manage asyncio event loop explicitly; understand run_until_complete
+- Never block I/O in async functions (no time.sleep(); use asyncio.sleep())
+- Set timeouts on all async operations using asyncio.wait_for()
+- Use asyncio.gather() or asyncio.TaskGroup for concurrent tasks
+- Understand GIL implications when using threading
+
+## Exception Handling
+- Never use bare except:; always catch specific exceptions
+- Use custom exception classes for domain errors
+- Use finally blocks for cleanup or context managers (with statement)
+- Log exceptions with context before re-raising
+- Don't silently swallow exceptions
+- Use context managers for all resource management
+
+## File I/O & Paths
+- Always use pathlib.Path instead of os.path
+- Use context managers for file operations: with open(...) as f:
+- Always specify encoding explicitly (utf-8 is typical)
+- Never hardcode absolute paths—use config or environment variables
+- Check file existence before operations using if path.exists()
+
+## Data Structures & Functions
+- Limit function length to ~30 lines; extract if larger
+- Limit function parameters to 3-4; use dataclass if more needed
+- Never use mutable defaults: def foo(items=[]) is wrong; use None + factory
+- Prefer dataclasses or Pydantic for structured data
+- Use enumerate() instead of range(len())
+- Use list/dict/set comprehensions for readability
+- Use generator expressions for large datasets
+
+## Testing & Documentation
+- Write docstrings for all public functions using Google style
+- Use pytest for unit tests; organize with fixtures and markers
+- Cover happy paths, edge cases, and error conditions
+- Use mocks/patches to isolate external dependencies
+- Never use print() in production—use logging module
+- Configure logging with proper levels: DEBUG, INFO, WARNING, ERROR
+
+## Performance & Best Practices
+- Avoid N+1 queries using select_related() or prefetch_related() in ORMs
+- Use pagination for large result sets
+- Cache expensive computations (consider functools.lru_cache)
+- Avoid algorithms with O(n²) or worse without justification
 - Profile before optimizing`,
-        go: `# Go Code Standards
-
-## Error Handling (Go Proverbs: "Errors are values")
-- Always handle every error; use _ = err only with explicit justification comment
-- Wrap errors with context: fmt.Errorf("doing X: %w", err)
-- Define custom error types for domain-specific errors
-- Never panic in libraries; only in main or tests
-- Provide helpful error messages to users
-- Use error chains to understand error context
-
-## Interface Design & Struct Composition
-- Keep interfaces small: 1-3 methods, defined where they're needed
-- Prefer composition over inheritance; use embedding for type reuse
-- Use pointer receivers when modifying receiver; value otherwise
-- Don't create interfaces for every type; only when needed
-- Use unexported (lowercase) fields by default; export only what's public
-- Document exported identifiers with comments starting with the name
-
-## Context & Concurrency
-- Pass context.Context as the first parameter in long-running functions
-- Always respect context cancellation in loops
-- Set timeouts on contexts: ctx, cancel := context.WithTimeout(ctx, time.Minute)
-- Never store context in struct fields; pass as parameter
-- Use context.Background() only at application entry point
-- Never block indefinitely—always have timeout or cancellation
-
-## Goroutines & Channel Safety
-- Prevent goroutine leaks by ensuring all goroutines exit
-- Use sync.WaitGroup or context cancellation to coordinate goroutines
-- Test with go run -race to detect data races
-- Always close channels from the sender; receivers don't close
-- Use select with timeout to prevent deadlocks
-- Protect shared memory with sync.Mutex or use channels
-
-## Code Style & Organization
-- Run gofmt automatically; never commit unformatted code
-- Run goimports to organize imports correctly
-- Keep function comments at the top, starting with function name
-- Document packages with a package comment at the top of the file
-- Group related code; use blank lines to separate logical blocks
-- File structure: package comment → imports → constants → vars → types → interfaces → functions
-
-## Function & File Structure
-- Place exported functions before unexported ones
-- Keep functions focused on a single task
-- Table-driven tests for testing multiple scenarios
-- Use subtests (t.Run) for test clarity and isolation
-- Max cyclomatic complexity ~5; refactor if exceeded
-- Package names: short, lowercase, no underscores
-
-## Testing Strategy
-- Use table-driven tests: test different inputs with same logic
-- Use t.Run for subtests and clearer test organization
-- Mock external dependencies; unit tests shouldn't call external services
-- Write benchmarks for performance-critical code
-- Fuzz tests for parsers and validators
-- Test error paths, not just happy paths
-
-## Common Go Patterns
-- Use defer for cleanup (files, locks, connections)
-- Check type assertions with v, ok := x.(Type)
-- Avoid init() functions; prefer explicit initialization
-- Use iota for enum-like constants
+        go: `# Go Code Standards
+
+## Error Handling (Go Proverbs: "Errors are values")
+- Always handle every error; use _ = err only with explicit justification comment
+- Wrap errors with context: fmt.Errorf("doing X: %w", err)
+- Define custom error types for domain-specific errors
+- Never panic in libraries; only in main or tests
+- Provide helpful error messages to users
+- Use error chains to understand error context
+
+## Interface Design & Struct Composition
+- Keep interfaces small: 1-3 methods, defined where they're needed
+- Prefer composition over inheritance; use embedding for type reuse
+- Use pointer receivers when modifying receiver; value otherwise
+- Don't create interfaces for every type; only when needed
+- Use unexported (lowercase) fields by default; export only what's public
+- Document exported identifiers with comments starting with the name
+
+## Context & Concurrency
+- Pass context.Context as the first parameter in long-running functions
+- Always respect context cancellation in loops
+- Set timeouts on contexts: ctx, cancel := context.WithTimeout(ctx, time.Minute)
+- Never store context in struct fields; pass as parameter
+- Use context.Background() only at application entry point
+- Never block indefinitely—always have timeout or cancellation
+
+## Goroutines & Channel Safety
+- Prevent goroutine leaks by ensuring all goroutines exit
+- Use sync.WaitGroup or context cancellation to coordinate goroutines
+- Test with go run -race to detect data races
+- Always close channels from the sender; receivers don't close
+- Use select with timeout to prevent deadlocks
+- Protect shared memory with sync.Mutex or use channels
+
+## Code Style & Organization
+- Run gofmt automatically; never commit unformatted code
+- Run goimports to organize imports correctly
+- Keep function comments at the top, starting with function name
+- Document packages with a package comment at the top of the file
+- Group related code; use blank lines to separate logical blocks
+- File structure: package comment → imports → constants → vars → types → interfaces → functions
+
+## Function & File Structure
+- Place exported functions before unexported ones
+- Keep functions focused on a single task
+- Table-driven tests for testing multiple scenarios
+- Use subtests (t.Run) for test clarity and isolation
+- Max cyclomatic complexity ~5; refactor if exceeded
+- Package names: short, lowercase, no underscores
+
+## Testing Strategy
+- Use table-driven tests: test different inputs with same logic
+- Use t.Run for subtests and clearer test organization
+- Mock external dependencies; unit tests shouldn't call external services
+- Write benchmarks for performance-critical code
+- Fuzz tests for parsers and validators
+- Test error paths, not just happy paths
+
+## Common Go Patterns
+- Use defer for cleanup (files, locks, connections)
+- Check type assertions with v, ok := x.(Type)
+- Avoid init() functions; prefer explicit initialization
+- Use iota for enum-like constants
 - Prefer if err != nil { return err } over else blocks`,
-        rust: `# Rust Code Standards
-
-## Ownership & Lifetimes
-- Understand and follow Rust's ownership rules strictly
-- Annotate lifetimes explicitly when the compiler can't infer
-- Use references (&T) for non-owning access; Box<T> for owned allocations
-- Pass &str instead of &String for function parameters
-- Pass &[T] instead of &Vec<T> for function parameters
-- Use &T for immutable access; &mut T for exclusive mutable access
-
-## Error Handling & Results
-- Use Result<T, E> for all fallible operations—never panic in libraries
-- Implement Error + Display traits for custom error types
-- Use the ? operator for error propagation
-- Add context to errors when wrapping
-- Never use unwrap() or expect() in production code (tests OK)
-- Consider custom error types instead of generic String errors
-
-## Smart Pointers & Memory Safety
-- Use Box<T> for owned heap allocations
-- Use Arc<T> for shared ownership; Rc<T> only in single-threaded contexts
-- Break circular references with Weak<T>
-- Understand Arc overhead and use judiciously
-- Never use raw pointers (*const T, *mut T) outside unsafe blocks
-- Never hold references across .await points without understanding
-
-## Unsafe Code
-- Minimize unsafe blocks; they should be last resort
-- Document safety invariants above every unsafe block
-- Include a SAFETY comment explaining why it's safe
-- Verify bounds before any pointer dereference
-- Use high-level safe abstractions instead of unsafe when possible
-- Review unsafe code with extra scrutiny during code review
-
-## Const Correctness & Generics
-- Use const for compile-time constant values
-- Use const fn for functions that can run at compile time
-- Use const generics for compile-time parameterization
-- Annotate mutable references only when mutation happens
-- Use &self for immutable methods; &mut self for mutating
-- Default to immutable bindings (let, not let mut)
-
-## Clippy & Code Quality
-- Enable #![warn(clippy::all)] or stricter in lib.rs/main.rs
-- Address all clippy warnings; understand before dismissing
-- Use cargo clippy to find idiomatic Rust patterns
-- Apply rustfmt for consistent formatting
-- Keep function length reasonable; extract if complex
-- DRY: refactor duplicated logic into helpers
-
-## Code Style & Documentation
-- Use snake_case for functions/variables, CamelCase for types/traits
-- Run cargo fmt before committing
-- Document all public functions with /// doc comments
-- Include examples in doc comments; they're compiled with cargo test --doc
-- Add SAFETY comments for all unsafe blocks explaining why it's safe
-- Write INVARIANT comments for complex data structures
-
-## Testing & Verification
-- Write unit tests in modules with #[cfg(test)]
-- Write integration tests in tests/ directory
-- Include examples in doc comments with /// \`\`\`
-- Test error paths and edge cases
-- Use property-based testing (proptest) for combinatorial coverage
-- Profile before optimizing; use cargo flamegraph
-
-## Trait & Generic Design
-- Keep traits cohesive; single responsibility
-- Use explicit generic bounds: fn foo<T: Clone + Display>(...)
-- Use associated types to avoid over-generalization
-- Follow orphan rule: implement foreign traits only on local types
-- Use HRTB (for<'a>) for advanced lifetime scenarios
+        rust: `# Rust Code Standards
+
+## Ownership & Lifetimes
+- Understand and follow Rust's ownership rules strictly
+- Annotate lifetimes explicitly when the compiler can't infer
+- Use references (&T) for non-owning access; Box<T> for owned allocations
+- Pass &str instead of &String for function parameters
+- Pass &[T] instead of &Vec<T> for function parameters
+- Use &T for immutable access; &mut T for exclusive mutable access
+
+## Error Handling & Results
+- Use Result<T, E> for all fallible operations—never panic in libraries
+- Implement Error + Display traits for custom error types
+- Use the ? operator for error propagation
+- Add context to errors when wrapping
+- Never use unwrap() or expect() in production code (tests OK)
+- Consider custom error types instead of generic String errors
+
+## Smart Pointers & Memory Safety
+- Use Box<T> for owned heap allocations
+- Use Arc<T> for shared ownership; Rc<T> only in single-threaded contexts
+- Break circular references with Weak<T>
+- Understand Arc overhead and use judiciously
+- Never use raw pointers (*const T, *mut T) outside unsafe blocks
+- Never hold references across .await points without understanding
+
+## Unsafe Code
+- Minimize unsafe blocks; they should be last resort
+- Document safety invariants above every unsafe block
+- Include a SAFETY comment explaining why it's safe
+- Verify bounds before any pointer dereference
+- Use high-level safe abstractions instead of unsafe when possible
+- Review unsafe code with extra scrutiny during code review
+
+## Const Correctness & Generics
+- Use const for compile-time constant values
+- Use const fn for functions that can run at compile time
+- Use const generics for compile-time parameterization
+- Annotate mutable references only when mutation happens
+- Use &self for immutable methods; &mut self for mutating
+- Default to immutable bindings (let, not let mut)
+
+## Clippy & Code Quality
+- Enable #![warn(clippy::all)] or stricter in lib.rs/main.rs
+- Address all clippy warnings; understand before dismissing
+- Use cargo clippy to find idiomatic Rust patterns
+- Apply rustfmt for consistent formatting
+- Keep function length reasonable; extract if complex
+- DRY: refactor duplicated logic into helpers
+
+## Code Style & Documentation
+- Use snake_case for functions/variables, CamelCase for types/traits
+- Run cargo fmt before committing
+- Document all public functions with /// doc comments
+- Include examples in doc comments; they're compiled with cargo test --doc
+- Add SAFETY comments for all unsafe blocks explaining why it's safe
+- Write INVARIANT comments for complex data structures
+
+## Testing & Verification
+- Write unit tests in modules with #[cfg(test)]
+- Write integration tests in tests/ directory
+- Include examples in doc comments with /// \`\`\`
+- Test error paths and edge cases
+- Use property-based testing (proptest) for combinatorial coverage
+- Profile before optimizing; use cargo flamegraph
+
+## Trait & Generic Design
+- Keep traits cohesive; single responsibility
+- Use explicit generic bounds: fn foo<T: Clone + Display>(...)
+- Use associated types to avoid over-generalization
+- Follow orphan rule: implement foreign traits only on local types
+- Use HRTB (for<'a>) for advanced lifetime scenarios
 - Prefer composition over trait objects when possible`,
-        java: `# Java Code Standards
-
-## Null Safety & Optional
-- Use Optional<T> instead of returning null
-- Use Optional.orElse(), orElseThrow(), orElseGet()
-- Never call Optional.get() without isPresent() check
-- Use @Nullable/@NonNull annotations for clarity
-- Handle NullPointerException in tests
-- Understand Optional is not a general-purpose wrapper
-
-## Resource Management (Try-With-Resources)
-- Always use try-with-resources for Closeable resources
-- Never use try-finally for resource cleanup
-- Connection/Statement always closed in try-with-resources
-- Stream/Reader/Writer closed in try-with-resources
-- Test for resource leaks with tools like NetBeans Profiler
-
-## Collections & Streams
-- Prefer Stream API over explicit loops for transformations
-- Use List.of(), Map.of(), Set.of() for immutable collections
-- Never create unnecessary new ArrayList/HashMap
-- Terminal operations required in Stream chains
-- Avoid nested flatMap; limit depth for readability
-- Use Collectors.groupingBy, toMap, etc. appropriately
-
-## Spring Boot & Dependency Injection
-- Constructor injection preferred over @Autowired on fields
-- Use @Autowired on constructor for single dependency
-- Avoid circular dependencies; refactor if found
-- Use @Service, @Repository, @Controller on appropriate classes
-- @Transactional on service methods, not on getters
-- Use @ConfigurationProperties for externalized configuration
-
-## JPA & ORM Patterns
-- Use fetch joins to prevent N+1 query problems
-- Avoid lazy loading issues; use eager loading when needed
-- Ensure entities only modified within transactions
-- Set @Transactional(readOnly=true) on query methods
-- Careful with cascade configuration—avoid unintended deletes
-- Use projections for read-only queries
-
-## Code Style & Conventions
-- 4-space indentation; never tabs
-- Max 100 characters per line
-- CamelCase for classes; camelCase for methods/variables
-- UPPER_SNAKE_CASE for constants
-- Always use @Override annotation
-- Comments explain "why", not "what"
-
-## Type Safety & Generics
-- Specify generic type bounds: <T extends SomeClass>
-- Use wildcard types appropriately: ? extends, ? super
-- Minimize unchecked casts; document when necessary
-- Never use raw types (List instead of List<String>)
-- Understand type erasure implications
-
-## Testing
-- Use JUnit 5 with @Test annotations
-- Mock external dependencies with Mockito
-- Test names describe what is tested
-- Follow Arrange-Act-Assert pattern
-- Document public APIs with Javadoc
-- Use @param, @return, @throws in Javadoc
-
-## Object-Oriented Design
-- Use inheritance sparingly; prefer composition
-- Mark immutable classes as final
-- Encapsulate fields; provide getters
-- Follow SOLID principles
-- Single Responsibility per class
+        java: `# Java Code Standards
+
+## Null Safety & Optional
+- Use Optional<T> instead of returning null
+- Use Optional.orElse(), orElseThrow(), orElseGet()
+- Never call Optional.get() without isPresent() check
+- Use @Nullable/@NonNull annotations for clarity
+- Handle NullPointerException in tests
+- Understand Optional is not a general-purpose wrapper
+
+## Resource Management (Try-With-Resources)
+- Always use try-with-resources for Closeable resources
+- Never use try-finally for resource cleanup
+- Connection/Statement always closed in try-with-resources
+- Stream/Reader/Writer closed in try-with-resources
+- Test for resource leaks with tools like NetBeans Profiler
+
+## Collections & Streams
+- Prefer Stream API over explicit loops for transformations
+- Use List.of(), Map.of(), Set.of() for immutable collections
+- Never create unnecessary new ArrayList/HashMap
+- Terminal operations required in Stream chains
+- Avoid nested flatMap; limit depth for readability
+- Use Collectors.groupingBy, toMap, etc. appropriately
+
+## Spring Boot & Dependency Injection
+- Constructor injection preferred over @Autowired on fields
+- Use @Autowired on constructor for single dependency
+- Avoid circular dependencies; refactor if found
+- Use @Service, @Repository, @Controller on appropriate classes
+- @Transactional on service methods, not on getters
+- Use @ConfigurationProperties for externalized configuration
+
+## JPA & ORM Patterns
+- Use fetch joins to prevent N+1 query problems
+- Avoid lazy loading issues; use eager loading when needed
+- Ensure entities only modified within transactions
+- Set @Transactional(readOnly=true) on query methods
+- Careful with cascade configuration—avoid unintended deletes
+- Use projections for read-only queries
+
+## Code Style & Conventions
+- 4-space indentation; never tabs
+- Max 100 characters per line
+- CamelCase for classes; camelCase for methods/variables
+- UPPER_SNAKE_CASE for constants
+- Always use @Override annotation
+- Comments explain "why", not "what"
+
+## Type Safety & Generics
+- Specify generic type bounds: <T extends SomeClass>
+- Use wildcard types appropriately: ? extends, ? super
+- Minimize unchecked casts; document when necessary
+- Never use raw types (List instead of List<String>)
+- Understand type erasure implications
+
+## Testing
+- Use JUnit 5 with @Test annotations
+- Mock external dependencies with Mockito
+- Test names describe what is tested
+- Follow Arrange-Act-Assert pattern
+- Document public APIs with Javadoc
+- Use @param, @return, @throws in Javadoc
+
+## Object-Oriented Design
+- Use inheritance sparingly; prefer composition
+- Mark immutable classes as final
+- Encapsulate fields; provide getters
+- Follow SOLID principles
+- Single Responsibility per class
 - No "god classes" with too many responsibilities`,
-        cpp: `# C++ Code Standards
-
-## Memory Management (RAII, Smart Pointers)
-- Use std::unique_ptr for single ownership
-- Use std::shared_ptr for shared ownership
-- Never use raw owning pointers (new/delete)
-- Use std::make_unique/make_shared
-- Ensure all resources have destructors
-- No memory leaks possible even in exception paths
-
-## Const Correctness
-- Mark methods const when they don't modify state
-- Use const references: const T& for parameters
-- Propagate const correctness up the call chain
-- Use mutable only for truly mutable implementation details
-- Mark data members const by default
-- const in templates: template<const T>
-
-## Pointer & Reference Safety
-- Use std::string_view for non-owning string parameters
-- Prefer references over pointers when nullability unnecessary
-- Use bounds checking in array operations
-- Check pointers before dereference
-- Use std::span<T> for array ranges (C++20)
-- Use std::optional<T> for optional values
-
-## RAII & Destructors
-- Every resource has a constructor (acquire) and destructor (release)
-- Define move constructor and move assignment
-- Copy constructor and assignment operator when deep copying needed
-- No resource leaks in exception paths
-- Use scopedexits pattern for cleanup
-- RAII applies to locks, files, network connections, memory
-
-## Standard Library
-- Use std::vector for dynamic arrays
-- Use std::string for text (not char arrays)
-- Use std::array for fixed-size arrays
-- Use std::map/unordered_map for key-value
-- Understand iterator validity before/after operations
-- Use range-based for loops: for(auto& x : container)
-- Use <algorithm> library functions
-
-## Error Handling
-- Exceptions preferred to error codes in modern C++
-- Provide strong or basic exception safety guarantee
-- Use noexcept appropriately; default to not noexcept
-- Custom exception types for domain errors
-- Avoid exception specifications; use noexcept only
-- Test exception paths in unit tests
-
-## Code Style & Quality
-- Run clang-format automatically; never commit unformatted code
-- Never use using namespace std (except in function scope)
-- Meaningful variable/function names
-- Single responsibility per function
-- Reasonable cyclomatic complexity
-- Named constants instead of magic numbers
-- Comments explain non-obvious logic
-
-## Testing & Documentation
-- Unit tests for critical functions
-- Integration tests for module interactions
-- Doxygen comments for public API
-- Include examples for complex APIs
-- README explains build and usage
-- Use Valgrind/AddressSanitizer for testing
-
-## Template Metaprogramming
-- Template complexity justified
-- Use SFINAE or C++20 concepts for overload resolution
-- Minimize template instantiation bloat
-- Document explicit instantiations
-- Use static_assert for compile-time checks
-- C++20 concepts preferred over enable_if
-
-## Performance
-- Use move semantics to avoid copies
-- Minimize allocations in hot loops
-- Reasonable algorithm complexity
-- Profile before optimizing
-- Cache-friendly data layout
+        cpp: `# C++ Code Standards
+
+## Memory Management (RAII, Smart Pointers)
+- Use std::unique_ptr for single ownership
+- Use std::shared_ptr for shared ownership
+- Never use raw owning pointers (new/delete)
+- Use std::make_unique/make_shared
+- Ensure all resources have destructors
+- No memory leaks possible even in exception paths
+
+## Const Correctness
+- Mark methods const when they don't modify state
+- Use const references: const T& for parameters
+- Propagate const correctness up the call chain
+- Use mutable only for truly mutable implementation details
+- Mark data members const by default
+- const in templates: template<const T>
+
+## Pointer & Reference Safety
+- Use std::string_view for non-owning string parameters
+- Prefer references over pointers when nullability unnecessary
+- Use bounds checking in array operations
+- Check pointers before dereference
+- Use std::span<T> for array ranges (C++20)
+- Use std::optional<T> for optional values
+
+## RAII & Destructors
+- Every resource has a constructor (acquire) and destructor (release)
+- Define move constructor and move assignment
+- Copy constructor and assignment operator when deep copying needed
+- No resource leaks in exception paths
+- Use scopedexits pattern for cleanup
+- RAII applies to locks, files, network connections, memory
+
+## Standard Library
+- Use std::vector for dynamic arrays
+- Use std::string for text (not char arrays)
+- Use std::array for fixed-size arrays
+- Use std::map/unordered_map for key-value
+- Understand iterator validity before/after operations
+- Use range-based for loops: for(auto& x : container)
+- Use <algorithm> library functions
+
+## Error Handling
+- Exceptions preferred to error codes in modern C++
+- Provide strong or basic exception safety guarantee
+- Use noexcept appropriately; default to not noexcept
+- Custom exception types for domain errors
+- Avoid exception specifications; use noexcept only
+- Test exception paths in unit tests
+
+## Code Style & Quality
+- Run clang-format automatically; never commit unformatted code
+- Never use using namespace std (except in function scope)
+- Meaningful variable/function names
+- Single responsibility per function
+- Reasonable cyclomatic complexity
+- Named constants instead of magic numbers
+- Comments explain non-obvious logic
+
+## Testing & Documentation
+- Unit tests for critical functions
+- Integration tests for module interactions
+- Doxygen comments for public API
+- Include examples for complex APIs
+- README explains build and usage
+- Use Valgrind/AddressSanitizer for testing
+
+## Template Metaprogramming
+- Template complexity justified
+- Use SFINAE or C++20 concepts for overload resolution
+- Minimize template instantiation bloat
+- Document explicit instantiations
+- Use static_assert for compile-time checks
+- C++20 concepts preferred over enable_if
+
+## Performance
+- Use move semantics to avoid copies
+- Minimize allocations in hot loops
+- Reasonable algorithm complexity
+- Profile before optimizing
+- Cache-friendly data layout
 - Inline hints used conservatively`,
-        kotlin: `# Kotlin Code Standards
-
-## Null Safety & Type System
-- Non-nullable types (T) preferred; nullable (T?) explicit
-- Safe calls (.?) and assertions (!!); assert only in tests
-- Elvis operator (?:) for defaults
-- Use val by default; var only when needed
-- lateinit for late initialization; lazy for computed values
-- Never use !!
-
-## Data Classes & Immutability
-- data class for value objects with auto-generated equals, hashCode, toString
-- Use copy() for "modifications" (maintains immutability)
-- val everywhere; var sparingly
-- Destructuring: val (x, y) = pair
-- Private constructors for controlled creation
-- sealed class for restricted hierarchies
-
-## Coroutines & Async
-- async/await instead of callbacks
-- Proper scope: GlobalScope avoided; use lifecycle scopes
-- Exception handling in coroutine builders
-- Cancellation respected in loops
-- launch vs async: launch for fire-and-forget, async for results
-- Job/Task properly awaited
-
-## Control Flow & When Expressions
-- when instead of if-else chains (>2 branches)
-- when exhaustive or explicit else
-- Scope functions (let, run, apply, also) appropriately used
-- if-expressions return values when possible
-- Early returns to avoid nesting
-- Guard clauses instead of nested blocks
-
-## Extension Functions & Higher-Order Functions
-- Extension functions for utility operations
-- Receiver clarity in extension scopes
-- No extensions on Any/Object
-- Lambdas single-expression when possible
-- it implicit parameter understood
-- Function types clear in signatures
-
-## Collections & Sequences
-- listOf, mapOf, setOf for immutable collections
-- Sequences for lazy evaluation chains
-- map/filter/reduce chains clear and efficient
-- groupBy for grouping; associate for map creation
-- No unnecessary list conversions
-- forEach/forEachIndexed over manual loops
-
-## Strings & Text
-- String templates (with ${'${}'}) instead of concatenation
-- Triple quotes (""" """) for multiline
-- Raw strings for regex/JSON
-- No string building in loops
-
-## Testing & Spring Integration
-- JUnit 5 with @Test
-- Mockk or similar for mocking
-- Test names describe behavior
-- Arrange-Act-Assert pattern
-- KDoc for public APIs
-- @SpringBootTest for integration tests
-
-## Code Style & Conventions
-- ktlint for consistent formatting
-- No redundant modifiers (public, final)
-- camelCase for functions; PascalCase for types
-- Comments explain "why" not "what"
-- Function length ~30 lines; file length ~300 lines
+        kotlin: `# Kotlin Code Standards
+
+## Null Safety & Type System
+- Non-nullable types (T) preferred; nullable (T?) explicit
+- Safe calls (.?) and assertions (!!); assert only in tests
+- Elvis operator (?:) for defaults
+- Use val by default; var only when needed
+- lateinit for late initialization; lazy for computed values
+- Never use !!
+
+## Data Classes & Immutability
+- data class for value objects with auto-generated equals, hashCode, toString
+- Use copy() for "modifications" (maintains immutability)
+- val everywhere; var sparingly
+- Destructuring: val (x, y) = pair
+- Private constructors for controlled creation
+- sealed class for restricted hierarchies
+
+## Coroutines & Async
+- async/await instead of callbacks
+- Proper scope: GlobalScope avoided; use lifecycle scopes
+- Exception handling in coroutine builders
+- Cancellation respected in loops
+- launch vs async: launch for fire-and-forget, async for results
+- Job/Task properly awaited
+
+## Control Flow & When Expressions
+- when instead of if-else chains (>2 branches)
+- when exhaustive or explicit else
+- Scope functions (let, run, apply, also) appropriately used
+- if-expressions return values when possible
+- Early returns to avoid nesting
+- Guard clauses instead of nested blocks
+
+## Extension Functions & Higher-Order Functions
+- Extension functions for utility operations
+- Receiver clarity in extension scopes
+- No extensions on Any/Object
+- Lambdas single-expression when possible
+- it implicit parameter understood
+- Function types clear in signatures
+
+## Collections & Sequences
+- listOf, mapOf, setOf for immutable collections
+- Sequences for lazy evaluation chains
+- map/filter/reduce chains clear and efficient
+- groupBy for grouping; associate for map creation
+- No unnecessary list conversions
+- forEach/forEachIndexed over manual loops
+
+## Strings & Text
+- String templates (with ${'${}'}) instead of concatenation
+- Triple quotes (""" """) for multiline
+- Raw strings for regex/JSON
+- No string building in loops
+
+## Testing & Spring Integration
+- JUnit 5 with @Test
+- Mockk or similar for mocking
+- Test names describe behavior
+- Arrange-Act-Assert pattern
+- KDoc for public APIs
+- @SpringBootTest for integration tests
+
+## Code Style & Conventions
+- ktlint for consistent formatting
+- No redundant modifiers (public, final)
+- camelCase for functions; PascalCase for types
+- Comments explain "why" not "what"
+- Function length ~30 lines; file length ~300 lines
 - Named arguments for functions with >3 parameters`,
-        php: `# PHP Code Standards
-
-## Type Safety & Strict Types
-- declare(strict_types=1) at the top of EVERY file
-- Type all parameters and return types
-- Type class properties
-- Avoid mixed type; be specific
-- Use typed properties (PHP 7.4+)
-- Static analysis with phpstan or psalm
-
-## Null Safety & Error Handling
-- Null coalescing (??) over isset checks
-- Nullsafe operator (?->) for deep property access
-- No silent null failures
-- Type hints include ? for nullable types
-- Guard clauses for early returns
-- Specific exception handling, not catch(Exception)
-
-## PSR Standards & Code Style
-- PSR-1: Basic Coding Standard
-- PSR-12: Extended Coding Style
-- PSR-4: Autoloading via Composer
-- 4-space indentation
-- CamelCase for classes; camelCase for methods
-- php_eol for line endings
-
-## Modern PHP Features (8.1+)
-- Enums for restricted value sets
-- Readonly properties for immutability
-- Match expressions instead of switch
-- Named arguments for clarity
-- Constructor promotion (public __construct(private string $name))
-- Fibers for async control flow
-
-## Eloquent & Database
-- Model relationships defined clearly
-- Eager loading (with()) prevents N+1
-- select() for specific columns
-- Query scopes for reusable logic
-- Soft deletes for historical data
-- Mass assignment protection ($guarded, $fillable)
-
-## Validation & Security
-- Request validation classes for input
-- Validate ALL user input
-- Custom validation messages
-- Gate/Policy for authorization
-- CSRF protection enabled
-- Parameterized queries (ORM handles this)
-
-## Collections & Functional
-- Illuminate\Support\Collection methods
-- map/filter/reduce for transformations
-- array functions for simple operations
-- No manual loops when collection methods suffice
-- Null-safe operations with collection methods
-- Type-safe collection usage
-
-## Laravel Patterns
-- Service/Repository for business logic
-- Middleware for cross-cutting concerns
-- Jobs/Queues for async operations
-- Service providers for registration
-- Facades appropriately used
-- Dependency injection via constructor
-
-## Testing
-- PHPUnit for unit tests
-- Feature tests for user workflows
-- Mocking external dependencies
-- Test database transactions
-- Meaningful assertion messages
-- Good test organization
-
-## Documentation & Strings
-- Docblock comments on public methods
-- README explains setup and usage
-- PHPDoc for parameter/return types
-- Heredoc/Nowdoc for multiline strings
-- No hardcoded values (use config)
+        php: `# PHP Code Standards
+
+## Type Safety & Strict Types
+- declare(strict_types=1) at the top of EVERY file
+- Type all parameters and return types
+- Type class properties
+- Avoid mixed type; be specific
+- Use typed properties (PHP 7.4+)
+- Static analysis with phpstan or psalm
+
+## Null Safety & Error Handling
+- Null coalescing (??) over isset checks
+- Nullsafe operator (?->) for deep property access
+- No silent null failures
+- Type hints include ? for nullable types
+- Guard clauses for early returns
+- Specific exception handling, not catch(Exception)
+
+## PSR Standards & Code Style
+- PSR-1: Basic Coding Standard
+- PSR-12: Extended Coding Style
+- PSR-4: Autoloading via Composer
+- 4-space indentation
+- CamelCase for classes; camelCase for methods
+- php_eol for line endings
+
+## Modern PHP Features (8.1+)
+- Enums for restricted value sets
+- Readonly properties for immutability
+- Match expressions instead of switch
+- Named arguments for clarity
+- Constructor promotion (public __construct(private string $name))
+- Fibers for async control flow
+
+## Eloquent & Database
+- Model relationships defined clearly
+- Eager loading (with()) prevents N+1
+- select() for specific columns
+- Query scopes for reusable logic
+- Soft deletes for historical data
+- Mass assignment protection ($guarded, $fillable)
+
+## Validation & Security
+- Request validation classes for input
+- Validate ALL user input
+- Custom validation messages
+- Gate/Policy for authorization
+- CSRF protection enabled
+- Parameterized queries (ORM handles this)
+
+## Collections & Functional
+- Illuminate\Support\Collection methods
+- map/filter/reduce for transformations
+- array functions for simple operations
+- No manual loops when collection methods suffice
+- Null-safe operations with collection methods
+- Type-safe collection usage
+
+## Laravel Patterns
+- Service/Repository for business logic
+- Middleware for cross-cutting concerns
+- Jobs/Queues for async operations
+- Service providers for registration
+- Facades appropriately used
+- Dependency injection via constructor
+
+## Testing
+- PHPUnit for unit tests
+- Feature tests for user workflows
+- Mocking external dependencies
+- Test database transactions
+- Meaningful assertion messages
+- Good test organization
+
+## Documentation & Strings
+- Docblock comments on public methods
+- README explains setup and usage
+- PHPDoc for parameter/return types
+- Heredoc/Nowdoc for multiline strings
+- No hardcoded values (use config)
 - Localization for user-facing strings`,
-        sql: `# SQL Code Standards
-
-## Security (SQL Injection Prevention)
-- Always use parameterized queries (? or :name)
-- Never string-interpolate user input
-- Input validation before SQL operations
-- User input never directly in WHERE/SELECT
-- Prepared statements cached and reused
-- Stored procedures with parameter binding
-- ORM systems with parameterization
-
-## Query Optimization
-- N+1 query prevention; use JOINs
-- SELECT specific columns, never SELECT *
-- Pagination for large result sets
-- EXPLAIN PLAN for complex queries
-- Indexes on WHERE, JOIN, ORDER BY columns
-- Join order matters for optimizer
-- Subqueries not in SELECT list
-
-## Schema Design & Constraints
-- Primary keys on every table
-- Foreign keys establish relationships
-- Data types match actual values
-- NOT NULL constraints explicit
-- UNIQUE constraints for natural keys
-- Normalization (3NF minimum)
-- No magic numbers as IDs
-
-## Migrations & Version Control
-- Migrations reversible (up/down)
-- Schema changes safe (add before remove)
-- Data migrations with rollback path
-- Test migrations up and down
-- Breaking changes documented
-- Backward compatibility maintained
-- One migration per change
-
-## Performance & Indexing
-- Indexes on high-cardinality columns
-- Composite indexes for multi-column filters
-- Covering indexes for hot queries
-- No redundant indexes
-- Index maintenance cost considered
-- Statistics updated regularly
-
-## Data Integrity & Testing
-- Audit columns (created_at, updated_at)
-- Soft deletes for historical data
-- Archival strategy documented
-- Backup/restore tested
-- Consistent validation rules
-- Temporal data handled correctly
-
-## Comments & Documentation
-- Comments on complex queries
-- Document table purposes
-- Explain non-obvious indexes
-- Record any known performance issues
+        sql: `# SQL Code Standards
+
+## Security (SQL Injection Prevention)
+- Always use parameterized queries (? or :name)
+- Never string-interpolate user input
+- Input validation before SQL operations
+- User input never directly in WHERE/SELECT
+- Prepared statements cached and reused
+- Stored procedures with parameter binding
+- ORM systems with parameterization
+
+## Query Optimization
+- N+1 query prevention; use JOINs
+- SELECT specific columns, never SELECT *
+- Pagination for large result sets
+- EXPLAIN PLAN for complex queries
+- Indexes on WHERE, JOIN, ORDER BY columns
+- Join order matters for optimizer
+- Subqueries not in SELECT list
+
+## Schema Design & Constraints
+- Primary keys on every table
+- Foreign keys establish relationships
+- Data types match actual values
+- NOT NULL constraints explicit
+- UNIQUE constraints for natural keys
+- Normalization (3NF minimum)
+- No magic numbers as IDs
+
+## Migrations & Version Control
+- Migrations reversible (up/down)
+- Schema changes safe (add before remove)
+- Data migrations with rollback path
+- Test migrations up and down
+- Breaking changes documented
+- Backward compatibility maintained
+- One migration per change
+
+## Performance & Indexing
+- Indexes on high-cardinality columns
+- Composite indexes for multi-column filters
+- Covering indexes for hot queries
+- No redundant indexes
+- Index maintenance cost considered
+- Statistics updated regularly
+
+## Data Integrity & Testing
+- Audit columns (created_at, updated_at)
+- Soft deletes for historical data
+- Archival strategy documented
+- Backup/restore tested
+- Consistent validation rules
+- Temporal data handled correctly
+
+## Comments & Documentation
+- Comments on complex queries
+- Document table purposes
+- Explain non-obvious indexes
+- Record any known performance issues
 - Database version requirements noted`,
-        csharp: `# C# Code Standards
-
-## Code Style & Conventions
-- Follow Microsoft C# Coding Conventions
-- PascalCase for public members; camelCase for private
-- 4-space indentation
-- Meaningful names for all identifiers
-- Max function length ~30 lines
-- Single Responsibility Principle
-
-## Language Features
-- LINQ for collection queries (not loops)
-- async/await for I/O operations
-- Nullable reference types (C# 8+)
-- Pattern matching in switch statements
-- Records for immutable data types
-- Tuples for returning multiple values
-- String interpolation ($"...")
-
-## Null Safety
-- Enable nullable reference types in csproj
-- Non-null types by default; ? for nullable
-- Use null-conditional operators (?., ?[])
-- Guard clauses for null checks
-- Null coalescing (??) for defaults
-
-## Dependency Injection & Architecture
-- Constructor injection for dependencies
-- Interface-based design
-- Dependency Injection containers
-- Inversion of Control pattern
-- SOLID principles
-
-## Testing & Documentation
-- Unit tests with xUnit or NUnit
-- Mock dependencies with Moq
-- Test names describe behavior
-- Integration tests for workflows
-- XML documentation on public APIs
-- Examples in documentation
-
-## async/await & Threading
-- async/await for I/O operations
-- ConfigureAwait(false) in libraries
-- Proper exception handling in async
-- Avoid sync-over-async patterns
-- Task.Run for CPU-bound work
-- Cancellation tokens passed through
-
-## Collections & Generics
-- LINQ over manual loops
-- Generic constraints explicit
-- Immutable collections for safety
-- IEnumerable for lazy evaluation
-- Choose collection type by use case
+        csharp: `# C# Code Standards
+
+## Code Style & Conventions
+- Follow Microsoft C# Coding Conventions
+- PascalCase for public members; camelCase for private
+- 4-space indentation
+- Meaningful names for all identifiers
+- Max function length ~30 lines
+- Single Responsibility Principle
+
+## Language Features
+- LINQ for collection queries (not loops)
+- async/await for I/O operations
+- Nullable reference types (C# 8+)
+- Pattern matching in switch statements
+- Records for immutable data types
+- Tuples for returning multiple values
+- String interpolation ($"...")
+
+## Null Safety
+- Enable nullable reference types in csproj
+- Non-null types by default; ? for nullable
+- Use null-conditional operators (?., ?[])
+- Guard clauses for null checks
+- Null coalescing (??) for defaults
+
+## Dependency Injection & Architecture
+- Constructor injection for dependencies
+- Interface-based design
+- Dependency Injection containers
+- Inversion of Control pattern
+- SOLID principles
+
+## Testing & Documentation
+- Unit tests with xUnit or NUnit
+- Mock dependencies with Moq
+- Test names describe behavior
+- Integration tests for workflows
+- XML documentation on public APIs
+- Examples in documentation
+
+## async/await & Threading
+- async/await for I/O operations
+- ConfigureAwait(false) in libraries
+- Proper exception handling in async
+- Avoid sync-over-async patterns
+- Task.Run for CPU-bound work
+- Cancellation tokens passed through
+
+## Collections & Generics
+- LINQ over manual loops
+- Generic constraints explicit
+- Immutable collections for safety
+- IEnumerable for lazy evaluation
+- Choose collection type by use case
 - No unnecessary allocations`,
     };
     return detailedRules[language] || BUILTIN_RULES[language] || `# Rules for ${language}\n(No rules defined for this language)`;
@@ -1031,7 +1031,7 @@ export function printRules() {
     for (const s of sets) {
         console.log(chalk.dim(`  ${s.language.padEnd(15)} [${s.source}]`));
     }
-    console.log(chalk.dim('  \n  Custom rules: ~/.crowcoder/rules/<language>.md'));
+    console.log(chalk.dim('  \n  Custom rules: ~/.compact-agent/rules/<language>.md'));
     console.log();
 }
 //# sourceMappingURL=rules.js.map