integration: Add PromptGuard AI security guardrails plugin

[acebot712]

Summary

Adds PromptGuard as a guardrails plugin for the Portkey AI Gateway. PromptGuard is an ML-powered AI security firewall that provides real-time threat detection and PII protection for LLM traffic.

Plugin functions

Function	Hook	Description
Security Scan	beforeRequestHook, afterRequestHook	ML-powered detection of prompt injection, jailbreaks, and toxicity
PII Redaction	beforeRequestHook, afterRequestHook	Detect and optionally redact PII (emails, phones, SSNs, credit cards, etc.)

Files added/changed

• plugins/promptguard/manifest.json — Plugin manifest with credentials and function definitions
• plugins/promptguard/scan.ts — Security scan handler (prompt injection, jailbreak, toxicity detection)
• plugins/promptguard/redact.ts — PII detection and redaction handler with content transformation
• plugins/promptguard/promptguard.test.ts — 7 unit tests covering credential validation, empty input, and edge cases
• conf.example.json — Added promptguard to plugins_enabled

Credentials

Field	Required	Description
apiKey	Yes	PromptGuard API key (encrypted). Get one at https://app.promptguard.co
baseUrl	No	Custom base URL for self-hosted deployments. Defaults to https://api.promptguard.co/api/v1

Configuration example

{
  "name": "promptguard-scan",
  "provider": "promptguard",
  "credentials": {
    "apiKey": "pg_live_..."
  },
  "function": "scan",
  "onSuccess": "allow",
  "onFail": "deny"
}

Test plan

• All 7 unit tests pass (npx jest plugins/promptguard/ --no-coverage)
• prettier --check passes (pre-commit hook verified)
• eslint passes with zero errors
• Build succeeds (rollup -c via pre-push hook)
• Gateway starts successfully (verified via pre-push hook)