| What it does | Deception-based AI security platform protecting AI systems with canary tokens, trap facts, prompt firewall, honeypots, and tamper-evident audit trails
| Make your software secure itself with autonomous security workers
|
|---|
| Stage | Not given
| Not given
|
|---|
| Founded | Not given
| Not given
|
|---|
| Based in | Not given
| Not given
|
|---|
| Pricing model | Freemium
| Freemium
|
|---|
| Pricing | Free plan with 3 decoy identities, 2 CanaryRAG documents, 10 Shield inspections per month, no credit card required. Shield at $12/mo includes 9 decoy identities, 3 protected emails, 500 Shield inspections per month. Pro at $24/mo includes everything in Shield plus tamper-evident logging at scale and
| Open source: free with unlimited usage. Private repositories: Free ($0), Starter ($499/month, 500 credits/month), Growth ($1,499/month, 2,000 credits/month), Scale (starts at $3,999/month, 6,000 credits/month). Annual plans include 20% more credits. Credits cost varies by security worker: PR Securit
|
|---|
| Key features | - Decoy email identities seeded into fake credential dumps on paste sites that alert when used
- Trap facts woven into documents to catch AI knowledge base leakage
- Three-layer prompt-injection firewall between users and models
- Fake AI endpoints that fingerprint attackers and classify their intent
- SHA-256 sealed, tamper-evident action logs for all agent actions
- OpenClaw agent heartbeat monitoring, skill scanning, and live action timeline
- Legal PDF export of tamper-evident records (Pro and above)
- API access on all plans including Free
- Email and Telegram alerts for all triggered canaries
- Threat-intel enrichment checking hits against IP reputation and exposure data
| - PR Security to stop risky code and untrusted contributors before they merge
- Secure Dependency Updates to turn vulnerable dependencies into tested, merge-ready pull requests
- Red Team to find exploit paths attackers would use before they do
- Agent Guardrails to let agents run without leaking secrets or breaking policy
- Security-One 27B decision model for always-on security triage
- Integration with existing tools and workflows without replacement
- Self-learning loop that improves security over time
- Protects applications, repositories, agents, and infrastructure
- API and MCP access across all plans
- Agent-to-agent remediation
|
|---|
| What makes it different | - Managed SaaS deployment with no hosting or maintenance required, unlike traditional honeypots
- Automated seeding of decoy credentials to paste platforms with staggered timing
- Purpose-built AI leak detection via trap facts in RAG systems
- Evidence-focused design with tamper-evident, verifiable logs
- Free plan with full access to all product surfaces for testing
| - Free for open source with unlimited usage across 3,000+ GitHub projects
- Security workers that each own one security job end to end
- Integrates seamlessly with existing tools without replacement needed
- Self-learning system that improves with each run
- Built by founders with 15+ years together and experience at Google, Klarna, and Trustly
|
|---|
| Community votes | 0 | 0 |
|---|
| DR (Domain Rating by Ahrefs) | Not checked yet ● no change since last check | 64 ● no change since last check |
|---|
| Trust Flow (Majestic) | 0 | 9 |
|---|
| Citation Flow (Majestic) | 0 | 40 |
|---|
| Referring domains (Majestic) | 0 | 609 |
|---|
CanaryVaults