Welcome to Hub Forge AI

You have received your credentials (username + password) by email or from your administrator.

• Open your browser and go to the Hub address
• Enter your username and your password
• You land on the Dashboard — your starting point

The Hub needs an AI engine to power the agents. Depending on your subscription, several options are available:

Option A — Your existing AI assistant (recommended):

• Hub Forge AI integrates with the AI tool you already use
• Go to Configuration → "AI Engine" section
• Select your tool: Claude Code, Cursor, GitHub Copilot, Windsurf or another
• Hub Forge AI orchestrates your agents through your own subscription — no additional AI cost
• Click Save

Option B — Direct cloud API:

• Go to Configuration → "AI Engine" section
• Select your provider: Anthropic (Claude), OpenAI (GPT), Google (Gemini) or Mistral
• Paste your API key in the dedicated field
• Click Save

Option C — Local AI (full sovereignty):

• Go to Configuration → "Local AI" section
• Compatible with: Ollama, LM Studio, vLLM, any OpenAI-compatible server
• Enter your server URL (e.g. http://localhost:11434/v1 for Ollama)
• Enter the model name (e.g. llama3.3:70b, mistral-large)
• Click Test connection then Save

Everything is configured. Launch your first mission:

• From the Dashboard, click Workspace
• You land on the workspace with the chat on the left and your missions on the right
• Type your first question in the input field at the bottom
• Examples: "Find me 5 cyber leads in Lyon" or "What are the NIS2 best practices?"
• Press Enter or click Send

The Hub is organized into 3 main areas:

• Workspace — Your daily workspace. Chat with the agent, manage your missions, leads table, audit log.
• Agent Monitor — Real-time supervision. Agent status, activity timeline, alerts, validation of critical actions. This is the view for the CISO or the administrator.
• Configuration — Settings for your environment. AI engine, MCP connectors, integrations.

The expertise centers (CyberForge, OpsForge, CompliForge, DataForge, LeadForge) are accessible from the Dashboard. Each has its own overview page.

The Workspace organizes your work into missions (independent chat sessions):

• New mission — Click "+ New" in the left column. Each mission has its own conversation history.
• Switch mission — Click a mission in the list to return to it.
• Purge a mission — The trash button at the top of the chat completely resets the session (history, context, limits).

The Hub protects your environment with two visible mechanisms:

• [SECURITY ALERT] — A message in red in the chat means suspicious external data has been blocked. The agent will not use it. This is the anti-injection shield (Pre-Hook) protecting you.
• [SYSTEM RATE LIMIT] — A message in orange means the agent tried to call the same tool too many times. It must change its approach. This is the FinOps safeguard that avoids costly loops.

On the Agent Monitor, critical actions (sending an email, writing to a system) appear in a red "Pending actions" banner. An administrator must approve them before execution.

The Workspace has 3 tabs at the top of the central area:

• Agent Chat — The conversation with the AI. This is where you work day to day.
• Leads — Table of your leads. Displays companies, decision-makers, emails and scores. Fills up as the agent generates Golden Records.
• Audit log — History of all critical actions (approved or rejected). Complete traceability for compliance.

The right-hand column displays the active context: current session, loaded project directives, change tracking, applied style, context optimization and active defense.