Nexus Shield LogoNexus Shield
Back to Home

Nexus Shield API Reference

Runtime security endpoints for Action Firewall, Inter-Agent Trust, and E2E demo scenarios.

Live DemoBenchmark Methodology
Authenticate all requests with x-api-key or x-nexus-api-key. Set BASE_URL to your dashboard origin or production host.
POST/api/v1/action/evaluate

Action Firewall & Intent Divergence

Evaluates a single agent tool call against declared capabilities, user intent, evidence chain, and divergence scoring.

cURL

curl -X POST "$BASE_URL/api/v1/action/evaluate" \
  -H "Content-Type: application/json" \
  -H "x-api-key: $NEXUS_API_KEY" \
  -d '{
    "agent_id": "crewai-ops-agent-1",
    "user_intent": "Check August Invoice 8291 for Acme Corp",
    "tool_call": { "name": "export_customer_database", "args": { "table": "customers" } },
    "agent_capabilities": ["READ", "DB_QUERY"]
  }'

Request

{
  "agent_id": "crewai-ops-agent-1",
  "user_intent": "Check August Invoice 8291 for Acme Corp",
  "tool_call": {
    "name": "export_customer_database",
    "args": {
      "table": "customers"
    }
  },
  "agent_capabilities": [
    "READ",
    "DB_QUERY"
  ]
}

Response

{
  "success": false,
  "decision": "BLOCK",
  "risk_score": 88,
  "intent_match_score": 35,
  "intent_divergence_percent": 96,
  "agent_status": "READ_ONLY",
  "capabilities_revoked": true,
  "violations": [
    "INTENT_ACTION_DIVERGENCE: 96% mismatch between user intent and action trajectory"
  ],
  "kill_switch_triggered": false,
  "latency_ms": 4.2
}

Try it out

POST/api/v1/agent/trust

Inter-Agent Trust Delegation

Verifies whether source agent may delegate work to target agent. Returns ALLOW_DELEGATION, REQUIRE_HUMAN_APPROVAL, or DENY_DELEGATION.

cURL

curl -X POST "$BASE_URL/api/v1/agent/trust" \
  -H "Content-Type: application/json" \
  -H "x-api-key: $NEXUS_API_KEY" \
  -d '{
    "source_agent_id": "langchain-support-agent-1",
    "target_agent_id": "crewai-ops-agent-1"
  }'

Request

{
  "source_agent_id": "langchain-support-agent-1",
  "target_agent_id": "crewai-ops-agent-1"
}

Response

{
  "success": true,
  "decision": "DENY_DELEGATION",
  "trust": {
    "sourceAgentId": "langchain-support-agent-1",
    "targetAgentId": "crewai-ops-agent-1",
    "trusted": false,
    "trustScore": 38,
    "targetReputation": 41,
    "recommendation": "DENY_DELEGATION",
    "rationale": [
      "Target agent has unresolved CRITICAL incidents"
    ]
  }
}

Try it out

GET/api/v1/agent/trust?agent_id=langchain-support-agent-1

Agent Reputation Card

Returns live Agent Reputation Card with trust score, risk badge, success rate, evidence verification ratio, and memory integrity score.

cURL

curl "$BASE_URL/api/v1/agent/trust?agent_id=langchain-support-agent-1" \
  -H "x-api-key: $NEXUS_API_KEY"

Response

{
  "success": true,
  "reputation_card": {
    "agentId": "langchain-support-agent-1",
    "reputationScore": 82,
    "riskBadge": "LOW",
    "metrics": {
      "successRate": 99,
      "blockedViolations": 1,
      "evidenceVerificationRatio": 100,
      "memoryIntegrityScore": 100
    }
  }
}

Try it out

POST/api/v1/demo/run

E2E Pitch Demo Scenario

Runs the investor-ready invoice → export → external upload mitigation scenario through the full SEE → CONTROL → TRUST engine stack.

cURL

curl -X POST "$BASE_URL/api/v1/demo/run" \
  -H "x-api-key: $NEXUS_API_KEY"

Response

{
  "success": true,
  "scenario": {
    "userIntent": "Check August Invoice 8291 for Acme Corp",
    "finalDivergenceScore": 96,
    "capabilityMode": "READ_ONLY",
    "evidenceStatus": "UNVERIFIED_ACTION",
    "reputationBefore": 92,
    "reputationAfter": 45
  }
}

Try it out