Open source · npm · about 8,000 users
Code Scanner
Run agent-security-scanner-mcp on a developer laptop. Scan code, prompts, packages, MCP servers, and tool configurations without a network call.
Download Open SourceProve your AI
is secure.
Continuously.
ProofLayer scans AI code before deployment, red-teams every agent continuously, and protects MCP traffic at runtime. Every finding becomes audit-ready evidence for SOC 2, NIST AI RMF, EU AI Act, and ISO 42001.
Trusted by security engineers at
Microsoft
Oracle
Uber
Zscaler
SUSE
Swiss Re
Schneider Electric
Truist
Samsung Fire & Marine Insurance
Euronext
Altinity
Alpaca
Clear Street
EMD Group (Merck KGaA)
Oak Ridge National Laboratory
Symplicity
V.tal
Ministero della Difesa
Aampe
Geordie AI
Netra Runtime
Predictive Labs
Rivoluzione Digitale
Silicon Interfaces
TechnoVal
VNO Design
Minas Code
Simplify (Macrotec)
SMC Global SecuritiesMicrosoftOracleUberZscalerSUSESwiss ReSchneider ElectricTruistSamsung Fire & Marine InsuranceEuronextAltinityAlpacaClear StreetEMD Group (Merck KGaA)Oak Ridge National LaboratorySymplicityV.talMinistero della DifesaAampeGeordie AINetra RuntimePredictive LabsRivoluzione DigitaleSilicon InterfacesTechnoValVNO DesignMinas CodeSimplify (Macrotec)SMC Global SecuritiesThe ProofLayer loopAttack. Detect. Prove.
Attack. Detect. Prove.
Then run it again.
Continuous attacks create current evidence. New models, tools, and agent workflows enter the same loop as you ship them.
AgentDojo benchmarkRed-team performance
Red-team performance
across real agent workflows.
Attack success and legitimate task utility measured across Workspace, Travel, Banking, and Slack suites.
Benchmark results
Attack success rate Utility
100.00%
89.46%
Workspace
560 cases100.00%
78.57%
Travel
140 cases100.00%
67.36%
Banking
144 cases100.00%
72.38%
Slack
105 cases100.00%
82.61%
Overall
949 casesUtility measures legitimate task completion while the attack policy runs.
About AgentDojoOne platform. Three product surfaces.Use the same detection engine
Use the same detection engine
from development to production.
Scan locally, attack on a schedule, and protect live MCP traffic. Every surface produces findings for one vulnerability report.
PyPI · LangChain and LangGraph
MCP Runtime Security
Place the shared detection engine inline with MCP traffic. Detect prompt injection, tool abuse, and data exfiltration during inference.
Download Open SourceScheduled batch service
Automated Red Teaming
Run nightly NEXUS campaigns across every deployed multi-agent system. Send verified exploits into the same vulnerability report.
Talk to usAttackTest the attack classes
Test the attack classes
your scanners miss.
Autonomous campaigns test prompt injection, jailbreaks, data exfiltration, tool abuse, RAG poisoning, and memory injection.
Prompt Injection
Override system instructions through direct and indirect injection — including system-prompt extraction and instruction hijacking across multi-turn conversations.
OWASP LLM01MITRE AML.T0051
Jailbreak
Bypass safety guardrails with DAN-style prompts, role-play exploits, and unrestricted-mode triggers tuned to the target model family.
OWASP LLM07
Exfiltration
Extract protected data via SQL-injection-through-LLM, secret exfiltration, and PII leakage. 289 verified exploits on multi-agent targets.
OWASP LLM06MITRE AML.T0024
289 verified exploits
Tool Abuse
Misuse agent tools: command injection, SSRF, path traversal, metadata extraction, and tool-call hijacking in MCP and LangChain agents.
OWASP LLM08
RAG Poisoning
Exploit document retrieval through semantic query manipulation, knowledge-base poisoning, and embedding-space attacks on vector stores.
OWASP LLM03
Memory Injection
13 attack families including false conversation history, temporal triggers, cross-session propagation, tool-description poisoning, and multi-agent function-call attacks.
OWASP LLM04MITRE AML.T0054
687 verified exploits · 13 families
976 verified exploits across 6 coordinated experts in one autonomous campaign.
ProveEvery attack becomes
Every attack becomes
audit-ready evidence.
Combine red-team findings with cloud and policy evidence. Generate packets for SOC 2, NIST AI RMF, EU AI Act, and ISO/IEC 42001.
SOC 2 Readiness
Cloud and policy evidence for the audit path.
Automated evidence collection across AWS, Azure, and GCP, Trust Services Criteria gap analysis, policy review, remediation roadmap, and a CPA-ready readiness packet.
AI Compliance Evidence Package
Governance artifacts for production AI systems.
Model inventory, AI risk register, governance documentation, and customer-facing transparency artifacts mapped to NIST AI RMF, EU AI Act, and ISO/IEC 42001.
AI Red-Team Assessment
Adversarial findings that become audit evidence.
Production agent workflows tested with established open-source tooling and ProofLayer's detection engine, with exploit scenarios mapped to OWASP LLM Top 10 and MITRE ATLAS.
One evidence trail
Audit artifacts mapped to the frameworks buyers already request.
Package verified red-team findings with the governance records needed by security teams, boards, and auditors.
Evidence artifacts
Model inventory
AI risk register
Governance policies
Transparency docs
Red-team findings
Executive briefing
Mapped frameworks
SOC 2
NIST AI RMF
EU AI Act
ISO/IEC 42001
OWASP LLM Top 10
MITRE ATLAS
Pricing
Community
Free
Open-source scanner for individual developers and small teams.
- Security scanner (CLI + MCP)
- 1,700+ detection rules
- Prompt injection probes
- MCP tool testing
- Community rule library
- MIT licensed
Enterprise
Custom
Dedicated red-teaming with compliance, SSO, and SLA guarantees.
- Everything in Community
- Continuous autonomous red-teaming
- Proof-of-exploit reports
- Compliance reporting (SOC 2, ISO 27001)
- SSO / SAML integration
- SLA guarantees
- CISO executive portal
- Dedicated support & onboarding
- Custom attack scenarios
Prove what happens
when your AI is attacked.
Run the attack. Replay the finding. Send the evidence.