Skip to content
SPECIFICTION
Get listed

PromptGuard

promptguard.co

From $19/mo Free plan WebWindowsmacOSLinuxAPISelf-hosted
Visit promptguard.co
Spec score 8.9 #2 of 31 in LLM Security Tools

What it is

PromptGuard is a security layer between an application and its LLM provider, inspecting requests before they reach the model. It describes 15 detectors across six layers for risks such as prompt injection, jailbreaks, PII, data exfiltration, toxicity, fraud, secrets, malware and tool injection. The product says it detects and redacts 43 PII types using reversible tokenization. Its SDK can automatically instrument supported LLM calls with one initialization call while leaving existing provider code unchanged. Listed providers include OpenAI, Anthropic, Gemini, Azure, Bedrock, Mistral, Cohere, DeepSeek, Groq, HuggingFace, Ollama and vLLM. Deployment options include managed cloud, hybrid self-hosting and air-gapped installations. In zero-retention mode, prompt and response content is not stored; default security events include a shortened preview and content hash. PromptGuard says customer prompts, completions and documents are not used to train, fine-tune or evaluate models. The Free tier is permanent and includes 20,000 scans a month, one API key, one project and 24-hour log retention. Paid plans include a 14-day money-back guarantee, not a trial.

Spec sheet

What it does
PromptGuard places a security gateway before an LLM provider to block prompt injection, jailbreaks, data exfiltration and PII leaks before model processing.Read 14 Oct 2026 · promptguard.co
Product
PromptGuard is a security layer between an application and its LLM provider that inspects requests before they reach the model.Read 30 Sep 2026 · promptguard.co
Deployment
The product offers managed cloud, hybrid self-hosting, and air-gapped deployment; air-gapped licences validate offline with a signed key.Read 30 Sep 2026 · promptguard.co
Trial
The pricing FAQ says Free is a permanent tier rather than a time-limited trial; paid plans include a 14-day money-back guarantee.Read 30 Sep 2026 · promptguard.co
Integrations
The site lists integrations/providers including OpenAI, Anthropic, Gemini, Azure, Bedrock, Mistral, Cohere, DeepSeek, Groq, HuggingFace, Ollama, and vLLM.Read 30 Sep 2026 · promptguard.co
Support
Pricing lists community support for Free, email support with a 48-hour response time for Pro, priority support with 24 hours for Scale, and dedicated support with four hours for Enterprise.Read 30 Sep 2026 · promptguard.co
Company
PromptGuard is the trading name of PG Tech Ltd, registered in England and Wales, with a registered office in London.Read 30 Sep 2026 · promptguard.co
AI-agent security
Agent security checks tool calls, privilege escalation and agent behavior before actions execute.Read 14 Oct 2026 · promptguard.co
Apple app pricing
The Apple listing is free with in-app purchases priced at $9.99 monthly or $79.99 annually for Pro, and $14.99 monthly or $114.99 annually for Family.Read 14 Oct 2026 · apps.apple.com
Apple app requirements
The Apple listing requires iOS or iPadOS 16, and macOS 13 on an Apple M1-or-later Mac.Read 14 Oct 2026 · apps.apple.com
Benchmark F1
PromptGuard reports an F1 score of 0.887 on 2,369 samples across eight datasets.Read 14 Oct 2026 · promptguard.co
Benchmark precision
The published benchmark reports 99.1% precision, with 10 false positives among 991 benign samples.Read 14 Oct 2026 · promptguard.co
Benchmark recall
The published benchmark reports 80.3% recall and says 271 of 1,378 attack samples were missed.Read 14 Oct 2026 · promptguard.co
Certifications
The trust page says SOC 2 Type II is not yet certified, ISO 27001 is planned, and GDPR and CCPA are supported.Read 30 Sep 2026 · promptguard.co
Chrome extension features
The listing describes live warnings, PII detection, one-click redaction, history scanning, a local dashboard, allowlists and custom keywords.Read 14 Oct 2026 · chromewebstore.google.com
Chrome extension privacy
The Chrome listing says detection runs locally and prompt or conversation text is not sent to the developer's servers.Read 14 Oct 2026 · chromewebstore.google.com
Company size and headquarters
LinkedIn lists PromptGuard as privately held, headquartered in London, with 2–10 employees.Read 14 Oct 2026 · linkedin.com
Data retention
Zero-retention mode avoids writing prompt and response bodies; metadata-only logs retain decisions, reasons and timestamps.Read 14 Oct 2026 · promptguard.co
Deployment modes
Customers can use managed cloud, hybrid self-hosting in their VPC, or a fully air-gapped installation.Read 14 Oct 2026 · promptguard.co
Documentation coverage
The documentation covers quickstart, APIs, SDKs, streaming, providers, CLI, VS Code, Cursor, GitHub scanning, MCP, dashboards, webhooks, audit logs, SSO, SCIM, self-hosting and reliability.Read 14 Oct 2026 · docs.promptguard.co
Failure behavior
The SDK defaults to fail-open when the Guard API is unavailable, with an option to fail-closed.Read 14 Oct 2026 · github.com
Hosted region
The hosted service runs on Google Cloud Run in us-central1 with managed Postgres, and no hosted EU region is offered.Read 14 Oct 2026 · promptguard.co
Integration methods
Developers can use auto-instrumentation, a direct Guard API, an HTTP proxy or a GitHub scanner.Read 14 Oct 2026 · promptguard.co
Latency
Deterministic fast-path checks are advertised as taking under 10 milliseconds, while escalated checks are network-bound.Read 14 Oct 2026 · promptguard.co
Legal maker
PromptGuard identifies PG Tech Ltd as its England-and-Wales trading company, registered at 128 City Road, London, EC1V 2NX.Read 14 Oct 2026 · promptguard.co
Limit: G2 enterprise depth
The sole G2 reviewer asks for more enterprise-grade capabilities, and G2 says there are too few reviews for buying insight.Read 14 Oct 2026 · g2.com
Limit: independent review false positives
AI Indigo reports false positives on complex legitimate prompts and says sensitivity tuning may be required.Read 14 Oct 2026 · aiindigo.com
Limit: independent review latency
AI Indigo measured roughly 0.15 seconds of added delay and warns that this could matter for ultra-low-latency or voice applications.Read 14 Oct 2026 · aiindigo.com
Notable limits
The product says five OWASP LLM Top 10 risks are covered in full and five are partial, with stated gaps including provenance verification and vector-store isolation.Read 30 Sep 2026 · promptguard.co
OWASP coverage limit
The company says five OWASP LLM Top 10 risks are covered fully and five only partially, with gaps including provenance, poisoning, vector-store isolation and spend caps.Read 14 Oct 2026 · promptguard.co
PII controls
PromptGuard says it detects and redacts 43 PII types, with reversible tokenization.Read 30 Sep 2026 · promptguard.co
PII protection
Its PII/DLP feature detects and redacts 43 entity types using reversible tokenization.Read 14 Oct 2026 · promptguard.co
Praise: G2 usability and value
The sole G2 reviewer praises quick setup, low cost, reliable support, broad LLM integration, monitoring and adjustable policies.Read 14 Oct 2026 · g2.com
Praise: independent review integration
AI Indigo praises context-aware PII redaction, drop-in middleware, provider flexibility and a dashboard with attack heat maps and alerts.Read 14 Oct 2026 · aiindigo.com
Provider integrations
The site lists OpenAI, Anthropic, Gemini, Azure, Bedrock, Mistral, Cohere, DeepSeek, Groq, HuggingFace, Ollama and vLLM compatibility.Read 14 Oct 2026 · promptguard.co
Public API
The Guard API accepts POST requests at /api/v1/guard and authenticates with an X-API-Key header.Read 14 Oct 2026 · docs.promptguard.co
Rating on AI Indigo
8.5 out of 10 from 1 reviewRead 14 Oct 2026 · aiindigo.com
Rating on Apple App Store
No rating displayed because the app has not received enough ratings or reviewsRead 14 Oct 2026 · apps.apple.com
Rating on Chrome Web Store
5 out of 5 from 5 reviewsRead 14 Oct 2026 · chromewebstore.google.com
Rating on G2
5 out of 5 from 1 reviewRead 14 Oct 2026 · g2.com
Red-team testing
The platform offers automated adversarial testing with more than 20 attack vectors and security grades.Read 14 Oct 2026 · promptguard.co
Reliability limitation
The stated 99.9% availability figure is an internal target rather than a signed self-serve SLA, and the company says it has no public status page.Read 14 Oct 2026 · promptguard.co
Residency
The hosted service runs on Google Cloud Run in us-central1, and the company says it does not currently offer a hosted EU region.Read 30 Sep 2026 · promptguard.co
SDK behavior
Its SDK can auto-instrument supported LLM calls with one initialization call while leaving existing provider code unchanged.Read 30 Sep 2026 · promptguard.co
SDK license
PromptGuard's published Python SDK is described as open source under the MIT license.Read 14 Oct 2026 · docs.promptguard.co
SDKs and frameworks
The Python SDK supports OpenAI, Anthropic, Google, Cohere and AWS Bedrock plus LangChain, CrewAI, LlamaIndex, AutoGen and Semantic Kernel integrations.Read 14 Oct 2026 · docs.promptguard.co
Security certifications
PromptGuard states that it is not SOC 2 Type II, ISO 27001 or FedRAMP certified and does not sign BAAs for hosted HIPAA use.Read 14 Oct 2026 · promptguard.co
Security compliance posture
The security page says GDPR and CCPA are supported, while EU AI Act and ISO/IEC 42001 controls are aligned.Read 14 Oct 2026 · promptguard.co
Security data handling
Zero-retention mode does not store prompt or response content, while default security events record a truncated 500-character preview and content hash.Read 30 Sep 2026 · promptguard.co
Self-host installation
Hybrid deployment is supplied as Helm or Docker Compose, and air-gapped licensing validates offline with signed keys.Read 14 Oct 2026 · promptguard.co
Threat detection
The product describes 15 detectors across six layers for threats including prompt injection, jailbreaks, PII, data exfiltration, toxicity, fraud, secrets, malware, and tool injection.Read 30 Sep 2026 · promptguard.co
Threat detectors
The product describes 15 detectors arranged in six layers, covering injection, jailbreaks, PII, exfiltration, toxicity, fraud, secrets, malware, URLs, tool injection, malicious entities, code vulnerabilities, gibberish and language policy.Read 14 Oct 2026 · promptguard.co
Training
PromptGuard says customer prompts, completions, and documents are never used to train, fine-tune, or evaluate models.Read 30 Sep 2026 · promptguard.co
Vulnerability response
PromptGuard says it acknowledges vulnerability reports within 48 hours and targets critical fixes within seven days.Read 14 Oct 2026 · promptguard.co

Plans and prices

Free Free 20,000 scans a month · 1 API key · 1 project · 24-hour log retention Read 30 Sep 2026 · promptguard.co
Free Free per month 20,000 scans/month · 1 API key · 1 project · 24-hour logs · requests pause at quota Read 14 Oct 2026 · promptguard.co
Team $19 a month 15,000 scans per seat, pooled · browser extension and macOS/Windows agent · fleet enrollment, MDM, org-wide policy Read 30 Sep 2026 · promptguard.co
Team $19 a month per seat/month 15,000 scans per seat pooled · browser extension · macOS/Windows agent · fleet enrollment · MDM · on-device secret blocking and PII redaction Read 14 Oct 2026 · promptguard.co
Build $99 a month monthly 100,000 scans/month · 5 API keys · 5 projects · full detector set · custom policies · email alerts Read 14 Oct 2026 · promptguard.co
Pro $99 a month 100,000 scans a month · 5 API keys · 5 projects · 7-day log retention Read 30 Sep 2026 · promptguard.co
Enterprise Priced on request custom contract custom volume · air-gapped deployment · SCIM · IP allowlist · custom retention · signed bundles · dedicated CSM · custom SLA Read 14 Oct 2026 · promptguard.co
Scale Price not published 500,000 requests/month · unlimited API keys and projects · 30-day log retention · overage metered at $0.40 per 1,000 requests up to spend cap Read 30 Sep 2026 · promptguard.co

Every PromptGuard plan and price →

Strong points and limits

Strong points

  • Detects and redacts 43 PII types.
  • Supports managed cloud, hybrid and air-gapped deployment.
  • SDK can instrument supported calls with one initialization.
  • Zero-retention mode does not store prompt or response content.

Limits

  • No free trial; paid plans have a money-back guarantee instead.
  • Hosted service does not offer an EU region.
  • SOC 2 Type II certification is not yet in place.
  • Five OWASP LLM Top 10 risks are only partially covered.

Who it suits

PromptGuard suits teams adding request inspection and data-protection controls to applications that use LLM providers. Its deployment options include self-hosted and air-gapped environments.

PromptGuard in depth

Read the full write-up

PromptGuard is an inspection layer for applications that send requests to large language models. It suits teams that need to screen LLM traffic and control sensitive data across cloud and self-managed deployments. Its breadth is useful, but incomplete risk coverage and US-only hosted residency rule it out for some requirements.

Overview

PromptGuard sits between an application and its LLM provider, inspecting requests before they reach the model. It describes 15 detectors across six layers, addressing threats such as prompt injection, jailbreaks, data exfiltration, PII, toxicity, fraud, secrets, malware and tool injection. The product also supports tests for prompt injection, jailbreaks, data leakage, unsafe outputs and custom cases, making it relevant to both live traffic screening and application checks.

Coverage is not comprehensive: PromptGuard says it covers five OWASP LLM Top 10 risks in full and five partially. Provenance verification and vector-store isolation are among the gaps, so teams with those controls as requirements should look elsewhere or plan additional safeguards.

Key features

Inspection and privacy controls

PromptGuard says it detects and redacts 43 PII types using reversible tokenization. That can help teams mask sensitive values while retaining the ability to restore them. Its SDK can auto-instrument supported LLM calls with one initialization call, leaving existing provider code unchanged; this is a practical advantage for teams adding inspection to an established application.

Zero-retention mode avoids storing prompt or response content. In the default mode, security events retain a truncated 500-character preview and a content hash, a meaningful distinction for organizations setting log policies. PromptGuard says customer prompts, completions and documents are never used to train, fine-tune or evaluate models.

Deployment and integrations

Managed cloud, hybrid self-hosting and air-gapped deployment accommodate different operational constraints. Air-gapped licences validate offline with a signed key, which is relevant to isolated environments. Integrations include OpenAI, Anthropic, Gemini, Azure, Bedrock, Mistral, Cohere, DeepSeek, Groq, HuggingFace, Ollama and vLLM, giving teams several provider choices without making the product itself a model provider.

The hosted service runs on Google Cloud Run in us-central1, and PromptGuard does not offer a hosted EU region. Organizations that require hosted EU residency should not assume self-hosted options resolve that requirement without checking their deployment needs. GDPR and CCPA are supported; SOC 2 Type II is not yet certified, and ISO 27001 is planned.

Pricing

PromptGuard uses a freemium model. The permanent Free plan costs nothing and includes 20,000 scans a month, one API key, one project and 24-hour log retention. With community support and a single project, it is a sensible evaluation tier, but its short retention and narrow project limit constrain ongoing multi-project use. There is no free trial; paid plans include a 14-day money-back guarantee.

Team costs $19 a month and provides 15,000 scans per seat, pooled, plus a browser extension, macOS and Windows agent, fleet enrollment, MDM and organization-wide policy. It is the entry paid choice for teams that need endpoint management and shared policy, though the quota is tied to seats rather than a single monthly allowance.

Pro costs $99 a month for 100,000 scans a month, five API keys, five projects and seven-day log retention. It fits teams running several integrations or projects that need more capacity than Free; the published plan facts do not describe Team's extension and fleet features as part of Pro.

Scale has custom pricing and includes 500,000 requests a month, unlimited API keys and projects, and 30-day log retention. Overage is metered at $0.40 per 1,000 requests up to a spend cap, so buyers should account for consumption beyond quota. Enterprise also has custom pricing, with custom volume, fully air-gapped deployment, SCIM, IP allowlist, custom retention and dedicated support with a four-hour response SLA. Support steps up from community on Free to email with a 48-hour response time on Pro, priority with 24 hours on Scale, and dedicated four-hour response on Enterprise.

Platforms

PromptGuard is offered across API, browser extension, Linux, macOS, self-hosted, web and Windows. That range supports both service integration and managed endpoint use, while hybrid and air-gapped deployment give security teams options beyond the hosted service.

Who it's for

PromptGuard is best suited to organizations building LLM applications that want request inspection, PII controls and deployment flexibility across multiple providers. Free can serve a small proof of concept; Team adds endpoint and fleet controls, while Pro raises scan and project capacity. Scale and Enterprise address higher volume, longer retention or tightly controlled deployment needs.

It is a weaker fit for buyers requiring complete coverage of the OWASP LLM Top 10, hosted EU residency, or an already-achieved SOC 2 Type II certification. Its own stated coverage gaps and trust posture make those requirements worth resolving before choosing a plan.

Pros and cons

Pros: Reversible tokenization across 43 PII types gives teams a specific privacy control rather than detection alone.

Pros: Managed, hybrid and air-gapped deployment, with offline licence validation, serve environments with different connectivity constraints.

Pros: Auto-instrumentation can add checks without changing supported provider code, reducing integration disruption.

Cons: Five OWASP LLM Top 10 risks are only partially covered, including gaps such as provenance verification and vector-store isolation.

Cons: Hosted service is in us-central1 with no hosted EU region, limiting it for residency-bound deployments.

Cons: SOC 2 Type II certification has not been achieved, which may disqualify it for some assurance requirements.

Alternatives

Consider LLM Security Tools, AI Guardrail Software or AI Security Testing Tools to compare options in those categories.

GuardionAI is a paid alternative with a free trial and an Enterprise plan offering unlimited requests and seats under custom contract pricing; consider it if those terms suit your procurement better.

Amazon Bedrock Guardrails is a paid, web-platform option with separately priced text content filters and denied topics, plus free image content filters; it may suit buyers comparing those specific controls and charges.

Openlayer Guardrails offers a free tier with 20,000 inference logs a month, 20 tests per project and three months of data retention, making it worth considering for test- and log-focused evaluation.

WitnessAI is a paid alternative with custom-priced Enterprise service and API, web and Windows platforms.

GLACIS offers a free account and API, self-hosted and web platforms, and may suit buyers comparing free-entry options.

Lasso AI Security Platform is a paid option with API, extension and web platforms.

HiddenLayer AI Guardrails is a paid, self-hosted and web option with a listed platform price of $5,000,000 a year on a 12-month contract; additional AWS infrastructure costs may apply.

Prisma AIRS AI Gateway uses consumption-based licensing through Software NGFW credits tied to token usage, with the required credits determined through sales.

Verdict

Choose PromptGuard if you need broad provider coverage, practical PII handling and a path from managed cloud to self-hosted or air-gapped deployment. Its strongest case is the combination of request inspection and deployment choice; look elsewhere if hosted EU residency, complete OWASP LLM Top 10 coverage or current SOC 2 Type II certification is non-negotiable.

Verdict

PromptGuard offers request inspection across many providers and deployment modes, with specific controls for PII and content retention. Consider the stated risk-coverage gaps and hosted region limits against your requirements.

Where Specifiction places it

Alternatives to PromptGuardAll alternatives →