diff --git a/common/src/templates/initial-agents-dir/skills/api-design-review/SKILL.md b/common/src/templates/initial-agents-dir/skills/api-design-review/SKILL.md
new file mode 100644
index 0000000000..7484051735
--- /dev/null
+++ b/common/src/templates/initial-agents-dir/skills/api-design-review/SKILL.md
@@ -0,0 +1,121 @@
+---
+name: api-design-review
+description: A structured guide for reviewing and designing RESTful and RPC-style APIs, covering endpoints, schemas, error handling, and best practices
+license: MIT
+metadata:
+  category: development
+  audience: developers
+---
+
+# API Design Review
+
+A structured guide for reviewing API designs, endpoints, and schemas.
+
+## When to use
+
+Use this skill when asked to review API endpoints, design new APIs, or evaluate API schemas and contracts.
+
+## Review Checklist
+
+### 1. Naming & Consistency
+
+- Are endpoint paths **plural nouns** (`/users`, `/orders`) not verbs (`/getUsers`, `/createOrder`)?
+- Are URL segments **kebab-case** or **snake_case** consistently?
+- Are query parameters consistently named (camelCase or snake_case)?
+- Do related resources follow a consistent pattern (`/users/:id/orders` not `/orders?userId=:id`)?
+- Is the API versioned (`/v1/`) in the URL path or header?
+
+### 2. RESTful Design
+
+- Are HTTP methods used correctly? (GET = read, POST = create, PUT = replace, PATCH = update, DELETE = remove)
+- Are status codes meaningful? (200 OK, 201 Created, 204 No Content, 400 Bad Request, 401 Unauthorized, 403 Forbidden, 404 Not Found, 409 Conflict, 422 Unprocessable, 500 Server Error)
+- Are responses consistent in structure across all endpoints?
+- Are list endpoints paginated with consistent params (`page`, `limit`, `cursor`)?
+- Are nested resources kept shallow (max 2-3 levels deep)?
+
+### 3. Request/Response Schemas
+
+- Are request bodies validated with consistent error messages?
+- Are response fields using **camelCase** for JSON APIs?
+- Are date/time fields using **ISO 8601** format?
+- Are IDs opaque strings (UUIDs) rather than sequential integers?
+- Are sensitive fields (passwords, tokens) excluded from responses?
+- Is there a consistent envelope for paginated responses (`{ data, meta: { total, page, limit } }`)?
+
+### 4. Error Handling
+
+- Is there a consistent error response format (`{ error: { code, message, details } }`)?
+- Are error messages human-readable and actionable?
+- Are internal error details (stack traces, SQL queries) hidden from responses?
+- Are rate limiting headers present (`X-RateLimit-Limit`, `X-RateLimit-Remaining`, `Retry-After`)?
+
+### 5. Security
+
+- Is authentication required for all endpoints unless explicitly public?
+- Are authorization checks performed per resource, not just at the endpoint level?
+- Are input sanitization and validation applied?
+- Is HTTPS enforced?
+- Are CORS headers properly configured?
+
+### 6. API Documentation
+
+- Is there an OpenAPI/Swagger specification?
+- Are there clear examples for each endpoint?
+- Are error scenarios documented?
+- Are deprecation policies documented?
+
+## Quick Reference
+
+```typescript
+// Good RESTful endpoint design
+GET    /v1/users              // List users (paginated)
+POST   /v1/users              // Create user
+GET    /v1/users/:id          // Get user by ID
+PATCH  /v1/users/:id          // Update user
+DELETE /v1/users/:id          // Delete user
+GET    /v1/users/:id/orders   // List user's orders (nested resource)
+
+// Consistent error format
+{
+  "error": {
+    "code": "VALIDATION_ERROR",
+    "message": "Email is required",
+    "details": {
+      "field": "email",
+      "constraint": "required"
+    }
+  }
+}
+
+// Consistent paginated response
+{
+  "data": [...],
+  "meta": {
+    "total": 100,
+    "page": 1,
+    "limit": 20,
+    "hasMore": true
+  }
+}
+```
+
+## Output Format
+
+Summarize findings as:
+
+- 🔴 **Breaking**: Incompatible change or security issue
+- 🟡 **Warning**: Should address before shipping
+- 🟢 **Suggestion**: Best practice improvement
+- ✅ **Praise**: Something done well
+
+## Quick Reference Commands
+
+- `read_files` — Read the API spec or endpoint files
+- `code_search` — Find related endpoints or usage patterns
+- `web_search` — Research API design standards or conventions
+
+## Notes
+
+- Design for your consumers, not your data model
+- Consistency beats cleverness every time
+- A good API is boring — predictable, simple, and hard to misuse
diff --git a/common/src/templates/initial-agents-dir/skills/code-review-checklist/SKILL.md b/common/src/templates/initial-agents-dir/skills/code-review-checklist/SKILL.md
new file mode 100644
index 0000000000..c591a969fb
--- /dev/null
+++ b/common/src/templates/initial-agents-dir/skills/code-review-checklist/SKILL.md
@@ -0,0 +1,98 @@
+---
+name: code-review-checklist
+description: A comprehensive code review checklist that guides agents through reviewing code changes for correctness, maintainability, performance, and security
+license: MIT
+metadata:
+  category: development
+  audience: developers
+---
+
+# Code Review Checklist
+
+A structured code review checklist for reviewing code changes systematically.
+
+## When to use
+
+Use this skill when asked to review code changes, perform a code review, or check for common issues in pull requests.
+
+## Review Dimensions
+
+### 1. Correctness
+
+- Do the changes actually solve the described problem?
+- Are edge cases handled (empty states, null values, error conditions)?
+- Are there any race conditions or timing issues?
+- Do the changes break existing functionality?
+
+### 2. Maintainability
+
+- Is the code readable and self-documenting?
+- Are functions and variables named clearly?
+- Is there unnecessary complexity that could be simplified?
+- Are there good comments explaining "why" (not "what")?
+- Would a new team member understand this code?
+
+### 3. Performance
+
+- Are there any obvious performance bottlenecks?
+- Are there unnecessary re-renders, recomputations, or network calls?
+- Are large data structures handled efficiently?
+- Is there proper memoization where appropriate?
+
+### 4. Security
+
+- Are user inputs properly validated and sanitized?
+- Are there any injection vulnerabilities (SQL, XSS, command injection)?
+- Are authentication/authorization checks in place?
+- Are secrets/hardcoded credentials exposed?
+- Are proper security headers or CSP policies applied?
+
+### 5. Testing
+
+- Are there tests for the new code?
+- Do existing tests still pass?
+- Are edge cases covered by tests?
+- Are test names descriptive?
+- Are there integration or e2e tests for critical paths?
+
+### 6. Architecture
+
+- Does the change follow the project's established patterns?
+- Are new dependencies justified?
+- Is the code placed in the right module/layer?
+- Does it introduce unnecessary coupling or circular dependencies?
+
+## Output Format
+
+Summarize findings as:
+
+- 🔴 **Critical**: Must fix before merging
+- 🟡 **Warning**: Should address soon
+- 🟢 **Suggestion**: Nice to have improvement
+- ✅ **Praise**: Something done well
+
+## Manual Testing Checklist (UI Changes)
+
+If review involves UI changes, also check:
+
+- Responsive/mobile layout
+- Loading states and skeletons
+- Error states and error messages
+- Empty states
+- Keyboard navigation and accessibility
+- Dark/light theme compatibility (if applicable)
+- Console errors or warnings
+
+## Quick Reference Commands
+
+When reviewing code, you can use these tools:
+
+- `read_files` — Read the changed files in full
+- `code_search` — Search for related patterns or usages
+- `run_terminal_command` — Run tests or linters
+
+## Notes
+
+- Be constructive, not critical. Suggest solutions, not just problems.
+- Prioritize findings: security > correctness > performance > maintainability > style
+- If unsure about a finding, mark it as a question rather than assuming the worst
diff --git a/common/src/templates/initial-agents-dir/skills/testing-strategies/SKILL.md b/common/src/templates/initial-agents-dir/skills/testing-strategies/SKILL.md
new file mode 100644
index 0000000000..1896902c9f
--- /dev/null
+++ b/common/src/templates/initial-agents-dir/skills/testing-strategies/SKILL.md
@@ -0,0 +1,150 @@
+---
+name: testing-strategies
+description: A practical guide for writing effective tests across unit, integration, and end-to-end levels, with patterns for TypeScript and React projects
+license: MIT
+metadata:
+  category: development
+  audience: developers
+---
+
+# Testing Strategies
+
+A practical guide for writing effective tests with the right scope, coverage, and patterns.
+
+## When to use
+
+Use this skill when asked to write tests, review test coverage, determine testing strategy, or improve test quality.
+
+## Testing Pyramid
+
+### 1. Unit Tests (fast, focused)
+
+- Test individual functions, classes, or modules in isolation
+- Mock external dependencies (APIs, databases, file system)
+- Cover: core logic, edge cases, error paths, boundary conditions
+- Goal: ~70% of your test suite
+
+### 2. Integration Tests (medium, connective)
+
+- Test how modules work together (API + database, service + repository)
+- Use real dependencies when practical, test doubles for external services
+- Cover: data flow, contract between layers, error propagation
+- Goal: ~20% of your test suite
+
+### 3. End-to-End Tests (slow, realistic)
+
+- Test complete user flows through the system
+- Use real infrastructure (test databases, staging APIs)
+- Cover: critical user journeys, deployment verification
+- Goal: ~10% of your test suite
+
+## Test Patterns
+
+### Arrange-Act-Assert
+
+Structure every test with clear phases:
+
+```typescript
+// Arrange
+const { result } = renderHook(() => useCounter())
+
+// Act
+act(() => result.current.increment())
+
+// Assert
+expect(result.current.count).toBe(1)
+```
+
+### Describe-It (Behavior-Driven)
+
+Name tests as specifications:
+
+```typescript
+describe('UserService', () => {
+  describe('getUser', () => {
+    it('returns user when found', async () => { ... })
+    it('throws NotFoundError when missing', async () => { ... })
+    it('caches result on repeated calls', async () => { ... })
+  })
+})
+```
+
+### Dependency Injection over Mocking
+
+Prefer passing dependencies rather than mocking:
+
+```typescript
+// Good: DI makes testing natural
+class OrderService {
+  constructor(
+    private db: Database,
+    private email: EmailService,
+  ) {}
+}
+
+// Test
+const service = new OrderService(mockDb, mockEmail)
+```
+
+### Test the Interface, Not Implementation
+
+- Test public API behavior, not internal details
+- Avoid testing private methods directly
+- Refactoring internals shouldn't break tests
+
+## What to Test
+
+### Always Test
+
+- Core business logic and calculations
+- Edge cases: empty states, null values, error conditions
+- Authentication and authorization
+- Data validation and sanitization
+- API error handling and status codes
+
+### Sometimes Test
+
+- UI component rendering (use snapshot tests sparingly)
+- Integration with third-party services (use contract tests)
+- Performance-critical paths
+
+### Don't Test
+
+- Generated code (scaffolds, ORM models)
+- Simple getters/setters
+- Language/stdlib features you didn't write
+- Configuration constants (test the code that uses them instead)
+
+## Quick Reference
+
+> **Note:** The examples below use `describe`/`test`/`expect` which work with Vitest, Jest, and Bun's built-in test runner (`bun:test`).
+
+```typescript
+// Unit test example (Bun/Vitest compatible)
+test('validateEmail rejects invalid emails', () => {
+  expect(validateEmail('not-an-email')).toBe(false)
+  expect(validateEmail('user@example.com')).toBe(true)
+  expect(validateEmail('')).toBe(false)
+  expect(validateEmail(null)).toBe(false)
+})
+
+// Async test with error
+test('getUser throws on not found', async () => {
+  await expect(
+    userService.getUser('nonexistent-id'),
+  ).rejects.toThrow('User not found')
+})
+```
+
+## Quick Reference Commands
+
+- `run_terminal_command` — Run tests with `bun test`, `npm test`, or `npx vitest run`
+- `read_files` — Read test files and source files
+- `code_search` — Find related test patterns or coverage gaps
+
+## Notes
+
+- Write tests in the same language as your production code
+- Keep tests simple — no conditional logic, loops, or complex abstractions
+- A failing test should tell you what's wrong from the error message alone
+- Run the specific test file, not the whole suite, during development