Platform/Agents

Agents are nodes. Drop one in.

Define an agent the same way you define a workflow — visually, conversationally, or both. Pick a model. Hand it MCP servers and published workflows as tools. Set memory, termination policy, and guardrails.

Composable definition

Five parts. Each one swappable.

Agent definition

research-agent.yaml

model:   claude-sonnet-4
tools:
  - mcp://confluence
  - mcp://salesforce
  - workflow://enrich-account
memory:  conversation + knowledge:product-docs
stop:    goal-met | max-steps:20 | human-veto
guard:
  cost-cap: $2.50 / run
  approval: [mcp://salesforce.write-*]
  rbac:    role:sales-eng
Model

Bring your own.

Claude, GPT, Gemini, open-weights via Bedrock or your own gateway. Swap per environment — Haiku in staging, Sonnet in production.

Tools

Drag in. Drag out.

Any MCP server. Any published Nexus workflow. The agent's tool list updates the moment the underlying server does.

Memory

Conversation + grounding.

Short-term turns plus retrieval over your knowledge base. Source ACLs travel with the chunk — agents can't surface what the user can't see.

Stop & guard

Pause when it matters.

Cost caps. Step limits. Mark sensitive tools approval-required. The agent halts; the operator portal lights up.

Observability

Every reasoning step. Every tool call. Replayable.

Trace

Step-by-step.

Reasoning, tool inputs, tool outputs — all captured per turn. Inspect any decision the agent made.

Replay

Rerun from any node.

Reproduce a failed run with the same inputs. Tweak a prompt, replay, compare. Same harness as workflows.

Compare

Two prompts side by side.

A/B prompts, models, or tool sets across the same input set. Aggregate cost, latency, and outcome diffs.

Build your first agent

From idea to running agent in a sitting.

We'll walk you through defining a research agent against your own MCP servers — in 30 minutes, on your data.