agentsclimarketplace

Convex security audit

Skill ulpi-io/plugin-marketplace/plugins/waynesutton/skills/convex-security-audit

Deep security review patterns for authorization logic, data access boundaries, action isolation, rate limiting, and protecting sensitive operationsFrom its SKILL.md

Install
npx -y skills add ulpi-io/plugin-marketplace --skill convex-security-audit

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

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

14.5 KB, ~3.6k tokens by cl100k_base, as published. Nobody here has run it

Convex Security Audit

Comprehensive security review patterns for Convex applications including authorization logic, data access boundaries, action isolation, rate limiting, and protecting sensitive operations.

Documentation Sources

Before implementing, do not assume; fetch the latest documentation:

Instructions

Security Audit Areas

  1. Authorization Logic - Who can do what
  2. Data Access Boundaries - What data users can see
  3. Action Isolation - Protecting external API calls
  4. Rate Limiting - Preventing abuse
  5. Sensitive Operations - Protecting critical functions

Authorization Logic Audit

Role-Based Access Control (RBAC)

// convex/lib/auth.ts
import { QueryCtx, MutationCtx } from "./_generated/server";
import { ConvexError } from "convex/values";
import { Doc } from "./_generated/dataModel";

type UserRole = "user" | "moderator" | "admin" | "superadmin";

const roleHierarchy: Record<UserRole, number> = {
  user: 0,
  moderator: 1,
  admin: 2,
  superadmin: 3,
};

export async function getUser(ctx: QueryCtx | MutationCtx): Promise<Doc<"users"> | null> {
  const identity = await ctx.auth.getUserIdentity();
  if (!identity) return null;
  
  return await ctx.db
    .query("users")
    .withIndex("by_tokenIdentifier", (q) => 
      q.eq("tokenIdentifier", identity.tokenIdentifier)
    )
    .unique();
}

export async function requireRole(
  ctx: QueryCtx | MutationCtx, 
  minRole: UserRole
): Promise<Doc<"users">> {
  const user = await getUser(ctx);
  
  if (!user) {
    throw new ConvexError({
      code: "UNAUTHENTICATED",
      message: "Authentication required",
    });
  }
  
  const userRoleLevel = roleHierarchy[user.role as UserRole] ?? 0;
  const requiredLevel = roleHierarchy[minRole];
  
  if (userRoleLevel < requiredLevel) {
    throw new ConvexError({
      code: "FORBIDDEN",
      message: `Role '${minRole}' or higher required`,
    });
  }
  
  return user;
}

// Permission-based check
type Permission = "read:users" | "write:users" | "delete:users" | "admin:system";

const rolePermissions: Record<UserRole, Permission[]> = {
  user: ["read:users"],
  moderator: ["read:users", "write:users"],
  admin: ["read:users", "write:users", "delete:users"],
  superadmin: ["read:users", "write:users", "delete:users", "admin:system"],
};

export async function requirePermission(
  ctx: QueryCtx | MutationCtx,
  permission: Permission
): Promise<Doc<"users">> {
  const user = await getUser(ctx);
  
  if (!user) {
    throw new ConvexError({ code: "UNAUTHENTICATED", message: "Authentication required" });
  }
  
  const userRole = user.role as UserRole;
  const permissions = rolePermissions[userRole] ?? [];
  
  if (!permissions.includes(permission)) {
    throw new ConvexError({
      code: "FORBIDDEN",
      message: `Permission '${permission}' required`,
    });
  }
  
  return user;
}

Data Access Boundaries Audit

// convex/data.ts
import { query, mutation } from "./_generated/server";
import { v } from "convex/values";
import { getUser, requireRole } from "./lib/auth";
import { ConvexError } from "convex/values";

// Audit: Users can only see their own data
export const getMyData = query({
  args: {},
  returns: v.array(v.object({
    _id: v.id("userData"),
    content: v.string(),
  })),
  handler: async (ctx) => {
    const user = await getUser(ctx);
    if (!user) return [];
    
    // SECURITY: Filter by userId
    return await ctx.db
      .query("userData")
      .withIndex("by_user", (q) => q.eq("userId", user._id))
      .collect();
  },
});

// Audit: Verify ownership before returning sensitive data
export const getSensitiveItem = query({
  args: { itemId: v.id("sensitiveItems") },
  returns: v.union(v.object({
    _id: v.id("sensitiveItems"),
    secret: v.string(),
  }), v.null()),
  handler: async (ctx, args) => {
    const user = await getUser(ctx);
    if (!user) return null;
    
    const item = await ctx.db.get(args.itemId);
    
    // SECURITY: Verify ownership
    if (!item || item.ownerId !== user._id) {
      return null; // Don't reveal if item exists
    }
    
    return item;
  },
});

// Audit: Shared resources with access list
export const getSharedDocument = query({
  args: { docId: v.id("documents") },
  returns: v.union(v.object({
    _id: v.id("documents"),
    content: v.string(),
    accessLevel: v.string(),
  }), v.null()),
  handler: async (ctx, args) => {
    const user = await getUser(ctx);
    const doc = await ctx.db.get(args.docId);
    
    if (!doc) return null;
    
    // Public documents
    if (doc.visibility === "public") {
      return { ...doc, accessLevel: "public" };
    }
    
    // Must be authenticated for non-public
    if (!user) return null;
    
    // Owner has full access
    if (doc.ownerId === user._id) {
      return { ...doc, accessLevel: "owner" };
    }
    
    // Check shared access
    const access = await ctx.db
      .query("documentAccess")
      .withIndex("by_doc_and_user", (q) => 
        q.eq("documentId", args.docId).eq("userId", user._id)
      )
      .unique();
    
    if (!access) return null;
    
    return { ...doc, accessLevel: access.level };
  },
});

Action Isolation Audit

// convex/actions.ts
"use node";

import { action, internalAction } from "./_generated/server";
import { v } from "convex/values";
import { api, internal } from "./_generated/api";
import { ConvexError } from "convex/values";

// SECURITY: Never expose API keys in responses
export const callExternalAPI = action({
  args: { query: v.string() },
  returns: v.object({ result: v.string() }),
  handler: async (ctx, args) => {
    // Verify user is authenticated
    const identity = await ctx.auth.getUserIdentity();
    if (!identity) {
      throw new ConvexError("Authentication required");
    }
    
    // Get API key from environment (not hardcoded)
    const apiKey = process.env.EXTERNAL_API_KEY;
    if (!apiKey) {
      throw new Error("API key not configured");
    }
    
    // Log usage for audit trail
    await ctx.runMutation(internal.audit.logAPICall, {
      userId: identity.tokenIdentifier,
      endpoint: "external-api",
      timestamp: Date.now(),
    });
    
    const response = await fetch("https://api.example.com/query", {
      method: "POST",
      headers: {
        "Authorization": `Bearer ${apiKey}`,
        "Content-Type": "application/json",
      },
      body: JSON.stringify({ query: args.query }),
    });
    
    if (!response.ok) {
      // Don't expose external API error details
      throw new ConvexError("External service unavailable");
    }
    
    const data = await response.json();
    
    // Sanitize response before returning
    return { result: sanitizeResponse(data) };
  },
});

// Internal action - not exposed to clients
export const _processPayment = internalAction({
  args: {
    userId: v.id("users"),
    amount: v.number(),
    paymentMethodId: v.string(),
  },
  returns: v.object({ success: v.boolean(), transactionId: v.optional(v.string()) }),
  handler: async (ctx, args) => {
    const stripeKey = process.env.STRIPE_SECRET_KEY;
    
    // Process payment with Stripe
    // This should NEVER be exposed as a public action
    
    return { success: true, transactionId: "txn_xxx" };
  },
});

Rate Limiting Audit

// convex/rateLimit.ts
import { mutation, query } from "./_generated/server";
import { v } from "convex/values";
import { ConvexError } from "convex/values";

const RATE_LIMITS = {
  message: { requests: 10, windowMs: 60000 }, // 10 per minute
  upload: { requests: 5, windowMs: 300000 },  // 5 per 5 minutes
  api: { requests: 100, windowMs: 3600000 },  // 100 per hour
};

export const checkRateLimit = mutation({
  args: {
    userId: v.string(),
    action: v.union(v.literal("message"), v.literal("upload"), v.literal("api")),
  },
  returns: v.object({ allowed: v.boolean(), retryAfter: v.optional(v.number()) }),
  handler: async (ctx, args) => {
    const limit = RATE_LIMITS[args.action];
    const now = Date.now();
    const windowStart = now - limit.windowMs;
    
    // Count requests in window
    const requests = await ctx.db
      .query("rateLimits")
      .withIndex("by_user_and_action", (q) => 
        q.eq("userId", args.userId).eq("action", args.action)
      )
      .filter((q) => q.gt(q.field("timestamp"), windowStart))
      .collect();
    
    if (requests.length >= limit.requests) {
      const oldestRequest = requests[0];
      const retryAfter = oldestRequest.timestamp + limit.windowMs - now;
      
      return { allowed: false, retryAfter };
    }
    
    // Record this request
    await ctx.db.insert("rateLimits", {
      userId: args.userId,
      action: args.action,
      timestamp: now,
    });
    
    return { allowed: true };
  },
});

// Use in mutations
export const sendMessage = mutation({
  args: { content: v.string() },
  returns: v.id("messages"),
  handler: async (ctx, args) => {
    const identity = await ctx.auth.getUserIdentity();
    if (!identity) throw new ConvexError("Authentication required");
    
    // Check rate limit
    const rateCheck = await checkRateLimit(ctx, {
      userId: identity.tokenIdentifier,
      action: "message",
    });
    
    if (!rateCheck.allowed) {
      throw new ConvexError({
        code: "RATE_LIMITED",
        message: `Too many requests. Try again in ${Math.ceil(rateCheck.retryAfter! / 1000)} seconds`,
      });
    }
    
    return await ctx.db.insert("messages", {
      content: args.content,
      authorId: identity.tokenIdentifier,
      createdAt: Date.now(),
    });
  },
});

Sensitive Operations Protection

// convex/admin.ts
import { mutation, internalMutation } from "./_generated/server";
import { v } from "convex/values";
import { requireRole, requirePermission } from "./lib/auth";
import { internal } from "./_generated/api";

// Two-factor confirmation for dangerous operations
export const deleteAllUserData = mutation({
  args: {
    userId: v.id("users"),
    confirmationCode: v.string(),
  },
  returns: v.null(),
  handler: async (ctx, args) => {
    // Require superadmin
    const admin = await requireRole(ctx, "superadmin");
    
    // Verify confirmation code
    const confirmation = await ctx.db
      .query("confirmations")
      .withIndex("by_admin_and_code", (q) => 
        q.eq("adminId", admin._id).eq("code", args.confirmationCode)
      )
      .filter((q) => q.gt(q.field("expiresAt"), Date.now()))
      .unique();
    
    if (!confirmation || confirmation.action !== "delete_user_data") {
      throw new ConvexError("Invalid or expired confirmation code");
    }
    
    // Delete confirmation to prevent reuse
    await ctx.db.delete(confirmation._id);
    
    // Schedule deletion (don't do it inline)
    await ctx.scheduler.runAfter(0, internal.admin._performDeletion, {
      userId: args.userId,
      requestedBy: admin._id,
    });
    
    // Audit log
    await ctx.db.insert("auditLogs", {
      action: "delete_user_data",
      targetUserId: args.userId,
      performedBy: admin._id,
      timestamp: Date.now(),
    });
    
    return null;
  },
});

// Generate confirmation code for sensitive action
export const requestDeletionConfirmation = mutation({
  args: { userId: v.id("users") },
  returns: v.string(),
  handler: async (ctx, args) => {
    const admin = await requireRole(ctx, "superadmin");
    
    const code = generateSecureCode();
    
    await ctx.db.insert("confirmations", {
      adminId: admin._id,
      code,
      action: "delete_user_data",
      targetUserId: args.userId,
      expiresAt: Date.now() + 5 * 60 * 1000, // 5 minutes
    });
    
    // In production, send code via secure channel (email, SMS)
    return code;
  },
});

Examples

Complete Audit Trail System

// convex/audit.ts
import { mutation, query, internalMutation } from "./_generated/server";
import { v } from "convex/values";
import { getUser, requireRole } from "./lib/auth";

const auditEventValidator = v.object({
  _id: v.id("auditLogs"),
  _creationTime: v.number(),
  action: v.string(),
  userId: v.optional(v.string()),
  resourceType: v.string(),
  resourceId: v.string(),
  details: v.optional(v.any()),
  ipAddress: v.optional(v.string()),
  timestamp: v.number(),
});

// Internal: Log audit event
export const logEvent = internalMutation({
  args: {
    action: v.string(),
    userId: v.optional(v.string()),
    resourceType: v.string(),
    resourceId: v.string(),
    details: v.optional(v.any()),
  },
  returns: v.id("auditLogs"),
  handler: async (ctx, args) => {
    return await ctx.db.insert("auditLogs", {
      ...args,
      timestamp: Date.now(),
    });
  },
});

// Admin: View audit logs
export const getAuditLogs = query({
  args: {
    resourceType: v.optional(v.string()),
    userId: v.optional(v.string()),
    limit: v.optional(v.number()),
  },
  returns: v.array(auditEventValidator),
  handler: async (ctx, args) => {
    await requireRole(ctx, "admin");
    
    let query = ctx.db.query("auditLogs");
    
    if (args.resourceType) {
      query = query.withIndex("by_resource_type", (q) => 
        q.eq("resourceType", args.resourceType)
      );
    }
    
    return await query
      .order("desc")
      .take(args.limit ?? 100);
  },
});

Best Practices

  • Never run npx convex deploy unless explicitly instructed
  • Never run any git commands unless explicitly instructed
  • Implement defense in depth (multiple security layers)
  • Log all sensitive operations for audit trails
  • Use confirmation codes for destructive actions
  • Rate limit all user-facing endpoints
  • Never expose internal API keys or errors
  • Review access patterns regularly

Common Pitfalls

  1. Single point of failure - Implement multiple auth checks
  2. Missing audit logs - Log all sensitive operations
  3. Trusting client data - Always validate server-side
  4. Exposing error details - Sanitize error messages
  5. No rate limiting - Always implement rate limits

References

What ships with it: 3 files

4.4 KB alongside SKILL.md

agents/

assets/

Gives 0 of the 12 instructions most audit compliance skills give in ~3.6k tokens

Counted across 937 of the 1,487 authors here whose files we hold, read 2026-08-07

  • Fetch latest guidelines before each reviewin 43 of 937, across 3 files
  • Group findings by severityin 43 of 937
  • Check files against all fetched rulesin 42 of 937, across 2 files
  • Output findings in terse file:line formatin 41 of 937, across 3 files
  • Ask user which files to review if none specifiedin 41 of 937, across 3 files
  • Read specified files or prompt user for filesin 39 of 937, across 1 file
  • Generate the audit reportin 33 of 937, across 30 files
  • Assign a severity to every findingin 25 of 937
  • Run automated accessibility scansin 23 of 937, across 13 files
  • Output a markdown audit reportin 22 of 937
  • Map findings to WCAG criteriain 20 of 937, across 10 files
  • Confirm audit scopein 19 of 937, across 9 files

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 326,852. 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.