PromptShield
Open source LLM security platform. Scans every prompt for PII and secrets, enforces policy, and routes to any provider. Runs entirely on your infrastructure.
Point your SDK at the gateway and it scans every prompt, enforces policy, and routes to your provider. No code changes to your app.
What it catches
- Secrets: AWS keys, GitHub tokens, OpenAI keys, Anthropic keys, Stripe keys, Slack tokens, DB connection strings, private keys, JWTs
- PII: email, phone, SSN, credit card, IBAN, passport, medical license (30+ types, 6 languages)
- Responses: optionally scans LLM output before it reaches your app
Block or mask per entity type. Policy is a YAML file — check it into git, hot-reloads without restart.
Three components
| Gateway only | + Engine | + Dashboard | |
|---|---|---|---|
| Provider routing | ✓ | ✓ | ✓ |
| Rate limiting | ✓ | ✓ | ✓ |
| Token budgets | ✓ | ✓ | ✓ |
| API key vault | ✓ | ✓ | ✓ |
| Audit logs | ✓ | ✓ | ✓ |
| Prometheus metrics | ✓ | ✓ | ✓ |
| Secret detection (150+ rules) | ✓ | ✓ | ✓ |
| PII detection | ✓ | ✓ | |
| Prompt injection detection | ✓ | ✓ | |
| Policy enforcement | ✓ | ✓ | |
| Response scanning | ✓ | ✓ | |
| Policy editor UI | ✓ | ||
| Audit log UI | ✓ | ||
| Key management UI | ✓ |
Only the gateway is required. Secret detection (Gitleaks, 150+ rules) is built in and runs without the engine. Add the engine for PII and injection detection, and the dashboard for a UI.
Policy actions
| Action | Behavior |
|---|---|
block | HTTP 403. LLM never called. Zero tokens consumed. |
mask | PII replaced with [ENTITY_TYPE]. Sanitized prompt forwarded. |
allow | Passes through unchanged. |
warn | Log and allow. Coming soon. |