Autonomous Agents & Assistants

Build, configure, and deploy autonomous AI agents capable of reasoning, calling tools, and querying external data sources.

• Portal Page: /projects/[projectId]/agents

graph TD
    Agent["Autonomous Agent (Support Bot)"]
    Agent --> LLM["Model & Parameters (claude-3-5-sonnet, Temp: 0.2)"]
    Agent --> Prompt["System Instructions & Persona"]
    Agent --> Skills["Attached Skills (Search KnowledgeBase)"]
    Agent --> Tools["Bound Tools & MCP Servers (SQL, GitHub)"]
    Agent --> Guard["Guardrail Policies (PII Redaction)"]

Agent Configuration Parameters #

1. System Prompt: Defines the core identity, reasoning guidelines, and behavioral boundaries for the agent.
2. Model Selection: Choose specific models or virtual tier aliases (e.g., smart-tier for reasoning).
3. Temperature & Top-P: Fine-tune determinism vs creativity.
4. Tool Attachments: Bind registered custom tools or external MCP server capabilities.
5. Skill Attachments: Attach structured few-shot prompt templates and operational skills.

Testing in the Playground #

Once configured, click Open in Playground to interactively test the agent's multi-turn reasoning and tool invocation trace.

Next Steps #

• Test agents interactively: Agent Playground
• Register custom tools: Tools & Working Flow
• Connect MCP servers: MCP Servers & Transports