Skip to content

Wizard Guide

The Forge wizard guides you through creating an agent in seven steps. Each step maps to a section of the agent YAML specification. You can navigate freely between steps — progress is saved automatically.

Define who your agent is.

FieldRequiredDescription
nameYesUnique slug identifier (e.g., sales-qualifier). Used in API calls.
display_nameNoHuman-readable name shown in dashboards.
descriptionNoWhat the agent does. Shown in the agent list and templates.
avatarNoURL or emoji for visual identification.
tagsNoSearchable labels for organizing agents.

Configure the LLM provider and parameters.

Primary model:

  • Select a provider (OpenAI, Anthropic, Ollama, etc.) and model.
  • Set parameters: temperature, top_p, max_tokens.

Fallback model (optional):

  • A secondary provider/model used when the primary is unavailable or hits its circuit breaker.

Routing strategy:

  • cost_optimized — Prefer the cheapest available provider.
  • latency_optimized — Prefer the fastest provider.
  • quality_first — Always use the primary unless it fails.
  • round_robin — Distribute requests evenly.

Select and configure the tools your agent can use.

  • Drag tools from the available tools panel on the left into the agent’s tool list.
  • Forge emits two tool types: builtin (a tool that ships in the node) and agent (a sub-agent, with the target in agent:). Tenant APIs, MCP servers and catalog integrations (api, mcp, integration) are edited in YAML — see Tenant APIs and MCP and Integrations.
  • For each tool, configure permissions and rate limits as needed.
  • Reorder tools by dragging them within the list.

The available tools are fetched from the connected Astromesh node via GET /v1/tools.

Choose how your agent processes requests.

Pattern:

PatternDescription
reactReason + Act loop. The model decides when to call tools and when to respond.
plan_and_executeCreate a plan first, then execute each step. Good for complex multi-step tasks.
parallel_fan_outRun multiple sub-tasks in parallel, then aggregate results.
pipelineSequential chain of processing steps.
supervisorA supervisor agent delegates to specialized sub-agents.
swarmPeer-to-peer agent collaboration.

Parameters:

  • max_iterations — Maximum number of reasoning/action cycles (default varies by pattern).
  • timeout_seconds — Hard timeout for the entire orchestration run.

Configure memory, guardrails, and permissions.

Select which memory types to enable:

  • Conversational — Chat history. Configure max_turns and strategy (sliding_window, summary, token_budget).
  • Semantic — Vector embeddings for long-term recall. Requires a configured vector backend.
  • Episodic — Event logs for tracking agent actions over time.

Forge offers the five rule types core 0.70 enforces, each on the side core accepts:

TypeSideActions (default first)Fields
pii_detectioninput, outputredact, block, warn
topic_filterinputblock, warnforbidden topics
max_lengthinputtruncate, block, warnmax_chars
content_filteroutputredact, block, warnpatterns / keywords
cost_limitinputblockmax_tokens_per_turn, count

Every rule takes an optional rejection message. A block returns a normal response with blocked: true. The security presets only produce valid rules (cost_limit at 4000). The same rules are edited in the canvas panel.

Define what the agent is allowed to do:

  • Allowed tools and tool categories.
  • Maximum cost per request.
  • Rate limits.

Write the system prompt that defines your agent’s behavior.

  • The editor supports Jinja2 template syntax for dynamic prompts.
  • Available template variables include agent identity fields, memory context, and custom variables.
  • Use the preview pane to see how the rendered prompt will look with sample data.
You are {{ agent.display_name }}, an AI assistant for {{ company_name }}.
Your role: {{ agent.description }}
{% if memory.summary %}
Previous conversation summary:
{{ memory.summary }}
{% endif %}

The final step shows a complete summary of your agent configuration.

  • YAML preview — The full generated agent YAML, ready for review.
  • Validation — Any configuration issues are highlighted with suggestions.
  • Target selection — Choose where to deploy: Local, Remote Node, or Nexus.
  • Click Deploy to create and activate the agent on the selected target.

After deployment, the agent appears in the Forge dashboard with status deployed and is immediately available via the API.

Built by FAiNANSU