agentsclimarketplace

Check access control model

Skill dykyi-roman/awesome-claude-code/skills/check-access-control-model

Analyzes PHP code for access control issues. Detects inline role checks, hardcoded permissions, mixed ACL/RBAC models, missing Voter/Policy pattern, and authorization logic in controllers.From its SKILL.md

Install
npx -y skills add dykyi-roman/awesome-claude-code --skill check-access-control-model

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

8.4 KB, as published. Nobody here has run it

Access Control Model Check

Analyze PHP code for access control anti-patterns that lead to inconsistent authorization, privilege escalation, and unmaintainable permission logic.

Detection Patterns

1. Inline Role Checks

<?php

declare(strict_types=1);

// BAD: Hardcoded role check scattered across code
final class ArticleController
{
    public function delete(Request $request, string $id): Response
    {
        if ($request->user()->role === 'admin') {
            $this->articleService->delete($id);
            return new Response(null, 204);
        }

        return new Response('Forbidden', 403);
    }
}

// BAD: String-based role comparison
if ($user->getRole() === 'editor' || $user->getRole() === 'admin') {
    // Allowed
}

// GOOD: Voter/Policy pattern
final class ArticleVoter extends Voter
{
    protected function supports(string $attribute, mixed $subject): bool
    {
        return $subject instanceof Article
            && in_array($attribute, ['VIEW', 'EDIT', 'DELETE'], true);
    }

    protected function voteOnAttribute(string $attribute, mixed $subject, TokenInterface $token): bool
    {
        $user = $token->getUser();

        return match ($attribute) {
            'DELETE' => $this->canDelete($user, $subject),
            'EDIT' => $this->canEdit($user, $subject),
            'VIEW' => true,
            default => false,
        };
    }

    private function canDelete(UserInterface $user, Article $article): bool
    {
        return $user->hasRole('ROLE_ADMIN')
            || $article->authorId()->equals($user->id());
    }

    private function canEdit(UserInterface $user, Article $article): bool
    {
        return $user->hasRole('ROLE_EDITOR')
            || $article->authorId()->equals($user->id());
    }
}

// Usage in controller
final class ArticleController
{
    public function delete(Request $request, string $id): Response
    {
        $article = $this->articleRepository->findOrFail(new ArticleId($id));
        $this->denyAccessUnlessGranted('DELETE', $article);

        $this->articleService->delete($article);

        return new Response(null, 204);
    }
}

2. Hardcoded Permission Strings

<?php

declare(strict_types=1);

// BAD: Magic permission strings everywhere
if ($user->hasPermission('manage_users')) { /* ... */ }
if ($user->hasPermission('edit_posts')) { /* ... */ }
if ($user->hasPermission('manage_users')) { /* ... */ } // Typo risk!

// GOOD: Permission enum
enum Permission: string
{
    case ManageUsers = 'manage_users';
    case EditPosts = 'edit_posts';
    case ViewReports = 'view_reports';
    case DeleteOrders = 'delete_orders';
}

// Usage
if ($user->hasPermission(Permission::ManageUsers)) { /* ... */ }

3. Mixed Authorization Models

<?php

declare(strict_types=1);

// BAD: Some endpoints use RBAC, others use ACL, others use nothing
final class UserController
{
    public function index(): Response
    {
        // RBAC-style
        if (!$this->security->isGranted('ROLE_ADMIN')) {
            throw new AccessDeniedHttpException();
        }
        return new Response($this->userService->list());
    }
}

final class OrderController
{
    public function show(string $id): Response
    {
        // ACL-style
        $order = $this->orderRepo->find($id);
        if ($order->userId() !== $this->getUser()->id()) {
            throw new AccessDeniedHttpException();
        }
        return new Response($order);
    }
}

final class ReportController
{
    public function index(): Response
    {
        // No authorization at all!
        return new Response($this->reportService->generate());
    }
}

// GOOD: Consistent Voter-based authorization across all controllers
// Each resource type has its own Voter
// All controllers use denyAccessUnlessGranted()

4. Authorization Logic in Controllers

<?php

declare(strict_types=1);

// BAD: Complex authorization logic in controller action
final class ProjectController
{
    public function update(Request $request, string $id): Response
    {
        $project = $this->projectRepo->find($id);
        $user = $request->user();

        // 15 lines of authorization logic in controller
        if ($user->role === 'admin') {
            // admin can do anything
        } elseif ($user->role === 'manager' && $project->teamId() === $user->teamId()) {
            // manager can edit own team projects
        } elseif ($project->ownerId() === $user->id()) {
            // owner can edit own project
        } else {
            return new Response('Forbidden', 403);
        }

        $this->projectService->update($project, $request->validated());

        return new Response($project);
    }
}

// GOOD: Authorization in middleware or Voter
#[IsGranted('EDIT', subject: 'project')]
final class ProjectController
{
    public function update(Request $request, #[MapEntity] Project $project): Response
    {
        $this->projectService->update($project, $request->validated());

        return new Response($project);
    }
}

5. Missing Deny-by-Default

<?php

declare(strict_types=1);

// BAD: Some routes have no authorization at all
// Routes file with no access control:
// Route::get('/admin/users', [AdminController::class, 'users']);
// Route::post('/admin/settings', [AdminController::class, 'updateSettings']);

// GOOD: Global middleware enforces deny-by-default
// security.yaml or middleware
// access_control:
//     - { path: ^/api/public, roles: PUBLIC_ACCESS }
//     - { path: ^/api, roles: IS_AUTHENTICATED_FULLY }
//     - { path: ^/admin, roles: ROLE_ADMIN }
//     - { path: ^/, roles: IS_AUTHENTICATED_FULLY }  # Deny-by-default

Grep Patterns

# Inline role checks (string comparison)
Grep: "->role\s*===\s*['\"]|->getRole\(\)\s*===\s*['\"]|role\s*==\s*['\"]" --glob "**/*.php"

# Hardcoded permission strings (not enum)
Grep: "hasPermission\(['\"]|can\(['\"]|isAllowed\(['\"]" --glob "**/*.php"

# Authorization checks in controllers
Grep: "->role|->getRole|isAdmin|isManager|hasRole" --glob "**/*Controller*.php"
Grep: "->role|->getRole|isAdmin|isManager|hasRole" --glob "**/*Action*.php"

# Voter/Policy pattern presence
Grep: "extends Voter|extends Policy|implements VoterInterface" --glob "**/*.php"

# Security attributes/annotations
Grep: "#\[IsGranted|@IsGranted|@Security|denyAccessUnlessGranted" --glob "**/*.php"

# Missing authorization (controllers without security)
Grep: "class.*Controller" --glob "**/*Controller*.php"
Grep: "class.*Action" --glob "**/*Action*.php"

# Permission enum existence
Grep: "enum.*Permission|enum.*Role" --glob "**/*.php"

Severity Classification

PatternSeverity
Missing deny-by-defaultπŸ”΄ Critical
No authorization on admin endpointsπŸ”΄ Critical
Inline role checks with string comparison🟠 Major
Authorization logic in controllers🟠 Major
Hardcoded permission strings (no enum)🟠 Major
Mixed RBAC/ACL models🟑 Minor
Missing Voter/Policy for resource access🟑 Minor

Output Format

### Access Control Issue: [Brief Description]

**Severity:** πŸ”΄/🟠/🟑
**Location:** `file.php:line`
**Type:** [Inline Check|Hardcoded Permission|Mixed Model|Controller Auth|No Auth]

**Issue:**
[Description of the access control anti-pattern]

**Risk:**
- Privilege escalation via inconsistent checks
- Forgotten authorization on new endpoints
- Permission logic impossible to audit

**Code:**
```php
// Problematic pattern

Fix:

// With proper access control

## When This Is Acceptable

- **Public API endpoints** -- Endpoints explicitly designed for unauthenticated access (e.g., login, registration, public data)
- **Internal microservice communication** -- Service-to-service calls behind network security where mutual TLS is used
- **CLI commands** -- Console commands that run with system-level privileges by design
- **Health check endpoints** -- /health, /ready, /live endpoints intended for load balancers

### False Positive Indicators
- Route is explicitly marked as public in security configuration
- Controller is behind a firewall rule that already enforces authentication
- Role check is inside a Voter or Policy class (correct location)

Keep looking

Skills are one crate of 627,801. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.