Rulepacks

Rulepacks are curated collections of security rules that you can apply to your PromptGuard projects. They provide pre-built detection patterns for specific threats, industries, or compliance requirements.

Overview

PromptGuard maintains two types of rulepacks:

Type	Description	Verification
Official	Created and maintained by the PromptGuard team	Cryptographically signed (Ed25519)
Community	Created by the community	Not signed

Official rulepacks are cryptographically signed using Ed25519 signatures. The signature is verified on every load to ensure the rulepack hasn’t been tampered with.

Endpoints

List Rulepacks

List all available rulepacks (official and community).

GET /api/v1/rulepacks

Headers:

X-API-Key: your_api_key

Response (200 OK)

{
  "official": [
    {
      "name": "prompt-injection-core",
      "version": "1.2.0",
      "description": "Core prompt injection detection patterns",
      "author": "PromptGuard",
      "categories": ["injection", "jailbreak"],
      "rule_count": 45,
      "is_signed": true
    },
    {
      "name": "pii-protection",
      "version": "1.0.0",
      "description": "PII detection and redaction patterns",
      "author": "PromptGuard",
      "categories": ["pii", "compliance"],
      "rule_count": 28,
      "is_signed": true
    }
  ],
  "community": [
    {
      "name": "healthcare-compliance",
      "version": "0.9.0",
      "description": "HIPAA-specific content filters",
      "author": "community-contributor",
      "categories": ["compliance", "healthcare"],
      "rule_count": 15,
      "is_signed": false
    }
  ]
}

Get Rulepack

Get a specific rulepack by name and version.

GET /api/v1/rulepacks/{name}?version=latest

Parameter	Type	In	Default	Description
`name`	`string`	Path	Required	Rulepack name
`version`	`string`	Query	`"latest"`	Version to retrieve

Response (200 OK)

{
  "metadata": {
    "name": "prompt-injection-core",
    "version": "1.2.0",
    "description": "Core prompt injection detection patterns",
    "author": "PromptGuard",
    "created_at": "2025-06-01T00:00:00Z",
    "min_engine_version": "1.0.0",
    "categories": ["injection", "jailbreak"]
  },
  "rules": [
    {
      "id": "pi-001",
      "name": "instruction_override",
      "description": "Detects 'ignore previous instructions' patterns",
      "pattern": "(?i)(ignore|disregard|forget)\\s+(all\\s+)?(previous|prior|above)\\s+(instructions|rules|guidelines)",
      "severity": "high",
      "confidence": 0.92,
      "enabled": true,
      "tags": ["injection", "instruction-override"]
    },
    {
      "id": "pi-002",
      "name": "role_impersonation",
      "description": "Detects attempts to impersonate system roles",
      "pattern": "(?i)(you are now|act as|pretend to be|you're now)\\s+(a|an|the)?\\s*(system|admin|root|developer)",
      "severity": "high",
      "confidence": 0.88,
      "enabled": true,
      "tags": ["injection", "role-impersonation"]
    }
  ],
  "rule_count": 45,
  "is_signed": true
}

Get Sample Rulepack

Get a sample rulepack for testing. No authentication required.

GET /api/v1/rulepacks/sample/core

Response (200 OK) Returns a sample rulepack with a few example rules — useful for understanding the rulepack structure before creating your own.

Rule Structure

Each rule in a rulepack contains:

Field	Type	Description
`id`	`string`	Unique rule identifier
`name`	`string`	Human-readable rule name
`description`	`string`	What this rule detects
`pattern`	`string`	Regex pattern for detection
`severity`	`string`	`"low"`, `"medium"`, `"high"`, or `"critical"`
`confidence`	`float`	Detection confidence (0.0 - 1.0)
`enabled`	`boolean`	Whether the rule is active
`tags`	`string[]`	Categorization tags

Rulepack Metadata

Field	Type	Description
`name`	`string`	Rulepack identifier
`version`	`string`	Semantic version (e.g., `"1.2.0"`)
`description`	`string`	What this rulepack protects against
`author`	`string`	Creator of the rulepack
`created_at`	`string`	ISO 8601 creation timestamp
`expires_at`	`string`	Optional expiration date
`min_engine_version`	`string`	Minimum PromptGuard engine version required
`categories`	`string[]`	Category tags

Signature Verification

Official rulepacks are signed using Ed25519 cryptographic signatures:

The rulepack content is serialized to a canonical JSON representation (sorted keys, deterministic)
The canonical data is signed with an Ed25519 private key
The signature is stored as Base64-encoded text
On load, the signature is verified against the public key to ensure integrity

This guarantees that official rulepacks haven’t been modified since they were published.

Code Examples

Python
Node.js
cURL

import requests
import os

api_key = os.environ.get("PROMPTGUARD_API_KEY")
headers = {"X-API-Key": api_key}

# List all rulepacks
response = requests.get(
    "https://api.promptguard.co/api/v1/rulepacks",
    headers=headers
)
rulepacks = response.json()

print("Official rulepacks:")
for rp in rulepacks["official"]:
    print(f"  {rp['name']} v{rp['version']} ({rp['rule_count']} rules)")

print("\nCommunity rulepacks:")
for rp in rulepacks["community"]:
    print(f"  {rp['name']} v{rp['version']} ({rp['rule_count']} rules)")

# Get a specific rulepack
response = requests.get(
    "https://api.promptguard.co/api/v1/rulepacks/prompt-injection-core",
    headers=headers,
    params={"version": "latest"}
)
rulepack = response.json()

print(f"\n{rulepack['metadata']['name']} v{rulepack['metadata']['version']}")
print(f"Signed: {rulepack['is_signed']}")
for rule in rulepack["rules"][:3]:
    print(f"  [{rule['severity']}] {rule['name']}: {rule['description']}")

const apiKey = process.env.PROMPTGUARD_API_KEY;
const headers = { 'X-API-Key': apiKey };

// List all rulepacks
const rulepacks = await fetch(
  'https://api.promptguard.co/api/v1/rulepacks',
  { headers }
).then(r => r.json());

console.log('Official rulepacks:');
rulepacks.official.forEach(rp => {
  console.log(`  ${rp.name} v${rp.version} (${rp.rule_count} rules)`);
});

// Get a specific rulepack
const rulepack = await fetch(
  'https://api.promptguard.co/api/v1/rulepacks/prompt-injection-core?version=latest',
  { headers }
).then(r => r.json());

console.log(`\n${rulepack.metadata.name} v${rulepack.metadata.version}`);
console.log(`Signed: ${rulepack.is_signed}`);

# List all rulepacks
curl https://api.promptguard.co/api/v1/rulepacks \
  -H "X-API-Key: $PROMPTGUARD_API_KEY"

# Get a specific rulepack
curl "https://api.promptguard.co/api/v1/rulepacks/prompt-injection-core?version=latest" \
  -H "X-API-Key: $PROMPTGUARD_API_KEY"

# Get sample rulepack (no auth required)
curl https://api.promptguard.co/api/v1/rulepacks/sample/core

Use Cases

Rulepack Category	Use Case
`injection`	Detect prompt injection and jailbreak attempts
`pii`	Identify and redact personally identifiable information
`compliance`	Industry-specific regulatory compliance (HIPAA, GDPR, SOC 2)
`toxicity`	Content moderation and safety filtering
`exfiltration`	Prevent data exfiltration through AI prompts

Getting Started

Advanced Features

API Reference

api-keys

presets

rulepacks

usage

projects

scrape

agent

Rulepacks

Rulepacks

Overview

Endpoints

List Rulepacks

Get Rulepack

Get Sample Rulepack

Rule Structure

Rulepack Metadata

Signature Verification

Code Examples

Use Cases

Next Steps

Security Overview

Policy Presets

Getting Started

Advanced Features

API Reference

api-keys

presets

rulepacks

usage

projects

scrape

agent

​Rulepacks

​Overview

​Endpoints

​List Rulepacks

​Get Rulepack

​Get Sample Rulepack

​Rule Structure

​Rulepack Metadata

​Signature Verification

​Code Examples

​Use Cases

​Next Steps

Security Overview

Policy Presets

Rulepacks

Overview

Endpoints

List Rulepacks

Get Rulepack

Get Sample Rulepack

Rule Structure

Rulepack Metadata

Signature Verification

Code Examples

Use Cases

Next Steps