Agentic Detection & Response

Detection & Response for what your AI agents actually do.

Nyx Security secures the agent estate on one spine. Agentic Posture maps where you're exposed. Agentic Detection & Response instruments every tool call, model invocation, and MCP action in-process — attributed to the identity and code path behind it — and stops the exploit in-path, in real time.

In-path, real-timeAttributed to code + identityMetadata, not contentSelf-host / BYOC
agent container
# ADR is in-process instrumentation — no code change
$ pip install nyx-monitor

NYX_ENABLED=1 NYX_ENDPOINT=https://nyx/api/ingest python agent.py
✓ nyx installed  boundaries=[model_client, tool_dispatch, mcp, effect]

The platform

One platform, two planes

Posture informs. Detection & Response detects and, where authorized, acts. Sell one plane to a depth buyer or both to a platform buyer — off the same agent-semantic spine.

Agentic Posture · SPM

Map the exposure

Read-only · out-of-band

Read-only visibility into your AI agent estate — what agents can reach, what an attacker would actually exploit, validated with closed-loop probing, without ever touching production.

read-onlyestate discoveryreachabilityclosed-loop probing
Agentic Detection & Response · ADR

Stop the exploit in-path

In-process · in-path · real-time

The first D&R category built on an agent-semantic data model — attributes every tool call, model invocation, and MCP action to the identity and code path behind it, with synchronous, in-path response.

in-processreal-time blockattributiontaint-aware

The one-way rule — architectural, not a slide: posture findings never become enforcement directives, and runtime telemetry feeds posture's risk model but never reaches back into the runtime path.

The problem

Agents are a substrate your D&R stack can't see

An AI agent reasons, calls tools, reaches the network, and runs code — an autonomous loop your EDR, SIEM, and AI gateways can't attribute or stop. One poisoned input can turn a helpful agent into an exfiltration tool.

Prompt injection → exfil

Untrusted content the agent fetches carries instructions it then acts on — reaching out to send your data somewhere it shouldn't.

Confused deputy

The agent wields real credentials and tools on behalf of an attacker who never had them.

Shadow credentials

Keys used at runtime that no inventory or scan ever saw — invisible to static posture.

Unbounded effects

Outbound HTTP, subprocesses, MCP tools — real-world side effects fired from a model's decision.

The category

D&R has the right motion. It was missing the agent.

Instrument the substrate, correlate identity + action + context, detect malicious behavior, respond in real time. That lineage is well-worn — until you reach the agent. A SIEM sees “a process opened a socket,” not “Agent X, invoked by identity Y, called tool Z with tainted input from source W.” That gap is the category ADR opens.

since ~2013

EDR

endpoint

Watches processes and syscalls on the host.

network

NDR

packets & flows

Watches traffic and flows across the wire.

estate

XDR

both + identity

Correlates endpoint and network, plus identity.

now

ADR

agent actions

Tool calls, model invocations, MCP calls, effect-layer writes — attributed to the identity and code path that triggered them.

How it works

An agent-semantic data model, captured in-process

ADR installs inside the agent — injected at interpreter startup with zero code change. It wraps the four boundaries where an agent meets the world, capturing intent and ground truth as native agent events, not retrofitted syscalls or netflows.

01 · model_client

Intent

The system prompt, messages, tools offered, and the tool_calls the model chose — OpenAI, Anthropic, Bedrock & Vertex.

02 · tool_dispatch

Action

Which agent ran which tool, with what input — LangChain, LlamaIndex, CrewAI & AutoGen.

03 · mcp

Protocol

Model Context Protocol calls, client and server side — tool, arguments, transport, and the authz assessment point.

04 · effect

Ground truth

The real side effects — outbound HTTP, subprocesses, egress — nested under the decision and identity that caused them.

Every event resolves to agent · identity · tool · args · taint · triggering line — attribution is the product, not a log line.

Two engines, one substrate

Policy where you want control. Autonomy where you can't write the rule.

The same instrumentation drives both halves of Detection & Response — deterministic control for known risk, autonomous detection for the behavior nobody wrote a rule for.

🛡
Deterministic · firewall-like

Policy engine

Written rules — allow, deny, redact, or hold on specific tools, arguments, identities, or taint sources. Start in shadow, promote per rule to enforce. Known-bad and known-risky patterns get a synchronous, in-path verdict before the effect lands.

allow / denyredactholdper-rule · fail-open
Autonomous · D&R

Anomaly engine

Behavioral baselining on agent action sequences — detects deviation from an agent's normal tool-use and invocation pattern with no human having written a rule for that specific deviation, and responds (hold, flag, or block, per policy) with no human in the loop at detection time.

baselinestaint trackingentity analyticsauto-respond

This is the structure mature D&R already has — IOC rules plus behavioral ML. A firewall alone can't autonomously respond to behavior nobody wrote a rule for. Nyx does both, on one substrate.

Differentiation

What ADR does that nothing else does

Legacy D&R has the right motion and the wrong data model. AI firewalls have a closer data model and the wrong motion. Nyx ADR is the first with both.

Capability Nyx ADR AI Firewalls / Gateways Legacy EDR / XDR CASB / API
Agent-semantic actions (tool calls, MCP, model invocations) prompt I/O only
In-process, in-path (not proxy / out-of-band) proxy/gateway wrong data model
Attributes action to triggering code path + identity process-level
Taint tracking across the invocation boundary
Synchronous real-time response (block / redact / hold) on content on host behavior rate-limit / revoke
Autonomous anomaly response (no pre-written rule) host / process

Security & trust

In the critical path, and does no harm

ADR is real-time and in-path — which means the monitor must never break, slow, or expose the system it protects. And it complements your SOC: distilled, sanitized findings flow downstream to the SIEM/XDR you already run.

Do no harm

Fail-open by design. If Nyx errors, the agent runs exactly as it would have. ~20µs of overhead per call.

Metadata, not content

Payloads are hashed at the source by default. Full-content capture is per-agent opt-in — never the default.

Secrets fingerprinted

Credentials are detected and fingerprinted for reconciliation — the secret itself is never recorded.

Your infrastructure

Deploy in your own cloud with row-level tenant isolation. BYOC and air-gapped self-hosting supported.

Get started

Start read-only. Enforce when you're ready.

Land with Agentic Posture — read-only, no trust cost — to see where the risk actually is. Add Detection & Response as the in-path layer once posture has shown you where. Shadow mode available; no architecture change to start seeing detections.

quickstart
$ pip install nyx-monitor
NYX_ENABLED=1 python your_agent.py