AGENT SKILL

mistral-enterprise-rbac

Configure Mistral AI enterprise access control and organization management. Use when implementing role-based permissions, managing team access, or setting up organization-level controls for Mistral AI. Trigger with phrases like "mistral access control", "mistral RBAC", "mistral enterprise", "mistral roles", "mistral permissions", "mistral team".

License 许可
MIT
最近更新
2026-11-11
Stars
★ 2.0k
Forks
231

01 前言 Front Matter

name
mistral-enterprise-rbac
author
Jeremy Longshore <jeremy@intentsolutions.io>
license
MIT
version
1.0.0
description
Configure Mistral AI enterprise access control and organization management. Use when implementing role-based permissions, managing team access, or setting up organization-level controls for Mistral AI. Trigger with phrases like "mistral access control", "mistral RBAC", "mistral enterprise", "mistral roles", "mistral permissions", "mistral team".
allowed-tools
Read, Write, Edit

03 SKILL.md 正文

Mistral AI Enterprise RBAC

Overview

Configure enterprise-grade access control for Mistral AI integrations within your organization.

Prerequisites

  • Mistral AI API access
  • Understanding of role-based access patterns
  • User management system in place
  • Audit logging infrastructure

Role Definitions

Role Permissions Use Case
Admin Full access, manage keys Platform administrators
Developer All models, full features Active development
Analyst Read-only, limited models Data analysis, reports
Service API access, specific model Automated systems
Viewer Read logs only Auditors, stakeholders

Instructions

Step 1: Define Permission Schema

// permissions.ts
export type MistralPermission =
  | 'chat:complete'
  | 'chat:stream'
  | 'embeddings:create'
  | 'models:list'
  | 'models:use:small'
  | 'models:use:large'
  | 'keys:manage'
  | 'usage:view'
  | 'audit:view';

export type MistralRole = 'admin' | 'developer' | 'analyst' | 'service' | 'viewer';

export const ROLE_PERMISSIONS: Record<MistralRole, MistralPermission[]> = {
  admin: [
    'chat:complete', 'chat:stream', 'embeddings:create',
    'models:list', 'models:use:small', 'models:use:large',
    'keys:manage', 'usage:view', 'audit:view',
  ],
  developer: [
    'chat:complete', 'chat:stream', 'embeddings:create',
    'models:list', 'models:use:small', 'models:use:large',
    'usage:view',
  ],
  analyst: [
    'chat:complete', 'embeddings:create',
    'models:list', 'models:use:small',
    'usage:view',
  ],
  service: [
    'chat:complete', 'chat:stream', 'embeddings:create',
    'models:use:small',
  ],
  viewer: [
    'models:list', 'usage:view', 'audit:view',
  ],
};

export function hasPermission(role: MistralRole, permission: MistralPermission): boolean {
  return ROLE_PERMISSIONS[role].includes(permission);
}

Step 2: User and Organization Management

interface MistralUser {
  id: string;
  email: string;
  role: MistralRole;
  organizationId: string;
  apiKeyId?: string;
  createdAt: Date;
  lastActiveAt?: Date;
}

interface MistralOrganization {
  id: string;
  name: string;
  plan: 'free' | 'pro' | 'enterprise';
  settings: {
    allowedModels: string[];
    maxRequestsPerDay: number;
    requireApproval: boolean;
  };
  createdAt: Date;
}

class OrganizationManager {
  async createOrganization(name: string, plan: MistralOrganization['plan']): Promise<MistralOrganization> {
    const org: MistralOrganization = {
      id: crypto.randomUUID(),
      name,
      plan,
      settings: this.getDefaultSettings(plan),
      createdAt: new Date(),
    };

    await db.organizations.insert(org);
    return org;
  }

  private getDefaultSettings(plan: MistralOrganization['plan']) {
    const settings = {
      free: {
        allowedModels: ['mistral-small-latest'],
        maxRequestsPerDay: 100,
        requireApproval: false,
      },
      pro: {
        allowedModels: ['mistral-small-latest', 'mistral-large-latest'],
        maxRequestsPerDay: 10000,
        requireApproval: false,
      },
      enterprise: {
        allowedModels: ['mistral-small-latest', 'mistral-large-latest', 'mistral-embed'],
        maxRequestsPerDay: 1000000,
        requireApproval: true,
      },
    };
    return settings[plan];
  }

  async addUser(orgId: string, email: string, role: MistralRole): Promise<MistralUser> {
    const user: MistralUser = {
      id: crypto.randomUUID(),
      email,
      role,
      organizationId: orgId,
      createdAt: new Date(),
    };

    await db.users.insert(user);
    return user;
  }

  async updateUserRole(userId: string, newRole: MistralRole): Promise<void> {
    await db.users.update({ id: userId }, { $set: { role: newRole } });

    // Audit log
    await auditLogger.log({
      action: 'user.role.updated',
      userId,
      newRole,
      performedBy: getCurrentUser().id,
    });
  }
}

Step 3: Permission Middleware

import { Request, Response, NextFunction } from 'express';

interface AuthenticatedRequest extends Request {
  user?: MistralUser;
  organization?: MistralOrganization;
}

function requirePermission(permission: MistralPermission) {
  return async (req: AuthenticatedRequest, res: Response, next: NextFunction) => {
    const user = req.user;

    if (!user) {
      return res.status(401).json({ error: 'Unauthorized' });
    }

    if (!hasPermission(user.role, permission)) {
      // Audit failed access attempt
      await auditLogger.log({
        action: 'permission.denied',
        userId: user.id,
        permission,
        resource: req.path,
      });

      return res.status(403).json({
        error: 'Forbidden',
        message: `Missing permission: ${permission}`,
      });
    }

    next();
  };
}

// Usage
app.post('/api/chat',
  requirePermission('chat:complete'),
  async (req: AuthenticatedRequest, res) => {
    // Handle chat request
  }
);

app.post('/api/chat/stream',
  requirePermission('chat:stream'),
  async (req: AuthenticatedRequest, res) => {
    // Handle streaming
  }
);

Step 4: Model Access Control

function requireModelAccess(model: string) {
  return async (req: AuthenticatedRequest, res: Response, next: NextFunction) => {
    const user = req.user!;
    const org = req.organization!;

    // Check organization allows this model
    if (!org.settings.allowedModels.includes(model)) {
      return res.status(403).json({
        error: 'Model not allowed',
        message: `Your organization does not have access to ${model}`,
      });
    }

    // Check user role allows this model
    const modelPermission = model.includes('large')
      ? 'models:use:large'
      : 'models:use:small';

    if (!hasPermission(user.role, modelPermission as MistralPermission)) {
      return res.status(403).json({
        error: 'Model access denied',
        message: `Your role does not allow access to ${model}`,
      });
    }

    next();
  };
}

Step 5: API Key Scoping

interface ScopedApiKey {
  id: string;
  key: string; // Hashed
  userId: string;
  organizationId: string;
  name: string;
  permissions: MistralPermission[];
  allowedModels: string[];
  rateLimit: {
    requestsPerMinute: number;
    tokensPerDay: number;
  };
  expiresAt?: Date;
  createdAt: Date;
}

class ApiKeyManager {
  async createScopedKey(
    userId: string,
    name: string,
    permissions: MistralPermission[],
    options?: {
      allowedModels?: string[];
      rateLimit?: Partial<ScopedApiKey['rateLimit']>;
      expiresInDays?: number;
    }
  ): Promise<{ id: string; key: string }> {
    const user = await db.users.findOne({ id: userId });
    if (!user) throw new Error('User not found');

    // Validate permissions don't exceed user's role
    for (const perm of permissions) {
      if (!hasPermission(user.role, perm)) {
        throw new Error(`Cannot grant permission ${perm} - exceeds user role`);
      }
    }

    const rawKey = `msk_${crypto.randomBytes(32).toString('hex')}`;
    const hashedKey = crypto.createHash('sha256').update(rawKey).digest('hex');

    const scopedKey: ScopedApiKey = {
      id: crypto.randomUUID(),
      key: hashedKey,
      userId,
      organizationId: user.organizationId,
      name,
      permissions,
      allowedModels: options?.allowedModels || ['mistral-small-latest'],
      rateLimit: {
        requestsPerMinute: options?.rateLimit?.requestsPerMinute || 60,
        tokensPerDay: options?.rateLimit?.tokensPerDay || 100000,
      },
      expiresAt: options?.expiresInDays
        ? new Date(Date.now() + options.expiresInDays * 24 * 60 * 60 * 1000)
        : undefined,
      createdAt: new Date(),
    };

    await db.apiKeys.insert(scopedKey);

    return { id: scopedKey.id, key: rawKey };
  }

  async validateKey(rawKey: string): Promise<ScopedApiKey | null> {
    const hashedKey = crypto.createHash('sha256').update(rawKey).digest('hex');
    const key = await db.apiKeys.findOne({ key: hashedKey });

    if (!key) return null;
    if (key.expiresAt && key.expiresAt < new Date()) return null;

    return key;
  }
}

Step 6: Usage Quotas

class UsageQuotaManager {
  async checkQuota(userId: string, orgId: string): Promise<{
    allowed: boolean;
    remaining: { requests: number; tokens: number };
    resetAt: Date;
  }> {
    const org = await db.organizations.findOne({ id: orgId });
    const today = new Date().toISOString().split('T')[0];

    const usage = await db.usage.findOne({
      organizationId: orgId,
      date: today,
    }) || { requests: 0, tokens: 0 };

    const maxRequests = org!.settings.maxRequestsPerDay;
    const remaining = {
      requests: Math.max(0, maxRequests - usage.requests),
      tokens: Math.max(0, 1000000 - usage.tokens), // Example limit
    };

    return {
      allowed: remaining.requests > 0 && remaining.tokens > 0,
      remaining,
      resetAt: new Date(new Date().setHours(24, 0, 0, 0)),
    };
  }

  async recordUsage(orgId: string, requests: number, tokens: number): Promise<void> {
    const today = new Date().toISOString().split('T')[0];

    await db.usage.updateOne(
      { organizationId: orgId, date: today },
      {
        $inc: { requests, tokens },
        $setOnInsert: { organizationId: orgId, date: today },
      },
      { upsert: true }
    );
  }
}

Output

  • Role definitions implemented
  • Permission middleware active
  • Model access control configured
  • API key scoping enabled

Error Handling

Issue Cause Solution
Permission denied Wrong role Update user role or permissions
Key expired TTL passed Generate new key
Quota exceeded Heavy usage Upgrade plan or wait for reset
Model not allowed Organization restriction Contact admin

Examples

Quick Permission Check

if (!hasPermission(user.role, 'chat:stream')) {
  throw new ForbiddenError('Streaming not allowed for your role');
}

Create Limited Service Key

const { key } = await keyManager.createScopedKey(
  serviceUserId,
  'background-processor',
  ['chat:complete'],
  {
    allowedModels: ['mistral-small-latest'],
    rateLimit: { requestsPerMinute: 10 },
    expiresInDays: 30,
  }
);

Resources

Next Steps

For major migrations, see mistral-migration-deep-dive.

04 同仓库更多 Skills

Dicklesworthstone
Dicklesworthstone/pi_agent_rust

web-artifacts-builder

Suite of tools for creating elaborate, multi-component claude.ai HTML artifacts using modern frontend web technologies (React, Tailwind CSS, shadcn/ui). Use for complex artifacts requiring state management, routing, or shadcn/ui components - not for simple single-file HTML/JSX artifacts.

前端开发
★ 2.0k
Dicklesworthstone
Dicklesworthstone/pi_agent_rust

webapp-testing

Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.

测试与质量前端开发
★ 2.0k
Dicklesworthstone
Dicklesworthstone/pi_agent_rust

neuropixels-analysis

Neuropixels neural recording analysis. Load SpikeGLX/OpenEphys data, preprocess, motion correction, Kilosort4 spike sorting, quality metrics, Allen/IBL curation, AI-assisted visual analysis, for Neuropixels 1.0/2.0 extracellular electrophysiology. Use when working with neural recordings, spike sorting, extracellular electrophysiology, or when the user mentions Neuropixels, SpikeGLX, Open Ephys, Kilosort, quality metrics, or unit curation.

★ 2.0k
Dicklesworthstone
Dicklesworthstone/pi_agent_rust

skill-creator

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.

Skill与智能体开发
★ 2.0k
Dicklesworthstone
Dicklesworthstone/pi_agent_rust

notebooklm

Use this skill to query your Google NotebookLM notebooks directly from Claude Code for source-grounded, citation-backed answers from Gemini. Browser automation, library management, persistent auth. Drastically reduced hallucinations through document-only responses.

Skill与智能体开发
★ 2.0k
Dicklesworthstone
Dicklesworthstone/pi_agent_rust

slack-gif-creator

Knowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack."

邮件与沟通图像与多媒体
★ 2.0k