Lindy Security Basics

Lindy Security Basics

Overview

Security practices for Lindy AI agents. Agents are autonomous — they connect to external services, execute actions, and handle data. Security focuses on: API key management, webhook authentication, agent permission scoping, integration account isolation, and connection sharing controls.

Prerequisites

• Lindy account with API access
• Understanding of which integrations your agents use
• For Enterprise: SSO/SCIM configuration access

Instructions

Step 1: API Key Management

# Store API key in environment variable — never in source code
export LINDY_API_KEY="lnd_live_xxxxxxxxxxxxxxxxxxxx"

# Or use a secret manager
# AWS Secrets Manager
aws secretsmanager create-secret \
  --name lindy/api-key \
  --secret-string "$LINDY_API_KEY"

# Google Secret Manager
echo -n "$LINDY_API_KEY" | gcloud secrets create lindy-api-key \
  --data-file=-

Key rotation schedule:

Step 2: Webhook Authentication

Every webhook trigger generates a unique secret key. Verify it on every inbound request:

// Webhook signature verification middleware
function verifyLindyWebhook(
  req: express.Request,
  res: express.Response,
  next: express.NextFunction
) {
  const authHeader = req.headers.authorization;
  const expectedToken = process.env.LINDY_WEBHOOK_SECRET;

  if (!authHeader || authHeader !== `Bearer ${expectedToken}`) {
    console.warn('Rejected unauthorized webhook attempt', {
      ip: req.ip,
      path: req.path,
      timestamp: new Date().toISOString(),
    });
    return res.status(401).json({ error: 'Unauthorized' });
  }

  next();
}

app.post('/lindy/callback', verifyLindyWebhook, (req, res) => {
  // Process verified webhook
  handleWebhook(req.body);
  res.json({ received: true });
});

Step 3: Agent Permission Scoping

Lindy agents access external services through authorized connections. Minimize blast radius:

Per-agent integration isolation:

• Authorize a dedicated Gmail account per agent (not your personal inbox)
• Create Slack bot tokens scoped to specific channels
• Use read-only database credentials where possible
• Create separate API keys for each integration

Connection sharing controls:

Step 4: Limit Agent Skill Surface Area

Agents with Agent Steps can choose which skills to use. Reduce risk:

• Start with 2-4 focused skills per agent (not the full catalog)
• Avoid giving agents both read AND write access to the same service unless necessary
• Separate "read" agents from "write" agents for critical systems
• Use conditions to gate destructive actions behind human approval

Step 5: Data Handling in Agents

Agent Prompt Security Patterns:

## Data Constraints
- Never include API keys, passwords, or tokens in responses
- Redact email addresses and phone numbers from summaries
- Do not forward customer data to channels outside #support
- If asked to perform an action outside your scope, respond:
  "I cannot perform that action. Please contact an admin."

Step 6: Audit Agent Activity

1. Task history: Review agent Tasks tab for unexpected actions
2. Integration access: Periodically review which services each agent can access
3. Credit anomalies: Sudden credit spikes may indicate misuse or misconfiguration
4. Connection review: Remove unused integrations from agents

Step 7: Enterprise Security Features

Available on Enterprise plan:

Security Checklist

• API keys stored in environment variables or secret manager
• .env file in .gitignore
• Webhook secrets generated and verified on every request
• Each agent uses minimum necessary integrations
• Separate integration credentials per agent where possible
• Agent prompts include data handling constraints
• Regular review of agent task history for anomalies
• Key rotation schedule defined and followed
• Enterprise: SSO enabled, SCIM configured

Error Handling

Resources

• Lindy Security
• Lindy Privacy Policy
• Lindy Documentation

Next Steps

Proceed to lindy-prod-checklist for production readiness.

Source: jeremylongshore/claude-code-plugins-plus-skills → skills/.curated/lindy-security-basics/SKILL.md

Also appears in: jeremylongshore/claude-code-plugins-plus-skills/plugins/saas-packs/lindy-pack/skills/lindy-security-basics/SKILL.md