Skip to content

Pluggedspace Console: User Guide & Workflows

Welcome to Pluggedspace Console, the central platform to configure, supervise, and scale your autonomous AI agent fleet. This guide details the step-by-step workflows for operating your dashboard and managing agents.

Quick Start: Workspace Setup

  1. Access the Portal: Navigate to https://agents.pluggedspace.org/login.
  2. Account Registration:
  3. Sign up with your Work Email, Password, and Organization Name (this establishes your secure, isolated workspace).
  4. Verify your email address via the confirmation link.
  5. Workspace Initialization: Your workspace automatically provisions your default agent fleet (Monica, Marketer, and Business Finder), along with widget API keys and access to the tool library.

Dashboard Navigation Overview

Section Description
Overview High-level summary of agent activity, system status, and pending items.
Inbox The Human-in-the-Loop queue for reviewing and approving high-risk agent actions.
Agent Fleet Configuration and management panels for pre-built agents (Monica, Marketer, Finder).
Custom Agents Visual builder for creating and deploying custom agents with specialized tools and skills.
Memory Bank Upload and manage company reference documents for semantic context and knowledge retrieval.
Marketplace Discover, install, rate, and publish community agent packages and templates.
Skills Library of reusable skills (e.g., WebsiteAuditor, InvoiceExtractor) attached to agents.
Integrations Connect third-party services (Stripe, Slack, Email, CRM, GitHub) via API keys and OAuth.
Policy & Guardrails Configure rules, approval thresholds, rate limits, and operational boundaries.
Command Center Live task telemetry, active process feeds, and emergency stop controls.
Artifacts Browse and download persistent files and reports generated by your agent fleet.
Settings Workspace preferences, API keys, domain whitelisting, and team member management.

Step-by-Step User Workflows

Workflow 1: Ingesting Knowledge into the Memory Bank

Agents deliver accurate, context-aware responses when provided with your business documentation.

  1. Navigate to Memory Bank from the sidebar.
  2. Drag and drop or upload your reference files (.pdf, .docx, .txt, .csv) — such as FAQs, service terms, product catalogs, or pricing sheets.
  3. Automatic Ingestion: The platform extracts the text and automatically indexes the knowledge.
  4. Once the status shows Ready (Green), all agents in your workspace can reference this knowledge when answering queries or executing tasks.

Workflow 2: Deploying Monica (Customer Support Chatbot)

Monica interacts directly with visitors on your website to answer questions and capture leads.

  1. Go to Agent Fleet > Monica.
  2. Select Autonomy Mode:
  3. Manual: Monica drafts responses, but all messages require human review.
  4. Hybrid (Recommended): Monica answers routine queries autonomously. Complex or sensitive queries (e.g., refund requests) are routed to the Inbox for approval.
  5. Autonomous: Monica resolves customer conversations automatically.
  6. Customize Personality: In the System Prompt field, specify tone and instructions (e.g., "You are a helpful and polite customer support specialist for Pluggedspace. Keep answers concise.").
  7. Embed Widget on Your Website:
  8. Copy the integration code snippet: html <script src="https://api.agents.pluggedspace.org/static/accounts/js/monica-widget.js?api_key=YOUR_API_KEY" async></script>
  9. Paste it before the closing </body> tag on your website.
  10. Whitelist your domain in Settings > Domain Locking to prevent unauthorized widget usage.

Workflow 3: Managing the Human Approval Queue (Inbox)

When agents operate in Hybrid or Manual mode, critical actions require human approval before execution.

  1. Navigate to Inbox in the sidebar.
  2. Select any pending action card to view details:
  3. Agent Name: The agent proposing the action (e.g., Monica, Finance Agent).
  4. Proposed Action: The specific operation (e.g., process_refund, send_campaign_email).
  5. Confidence Score: AI self-assessed confidence level (e.g., 94%).
  6. Risk Level: Low, Medium, or High.
  7. AI Reasoning: Explanation of why the agent chose this action.
  8. Click Approve to execute the action immediately, or Reject to cancel it.

Workflow 4: Connecting Third-Party Integrations

Empower agents to perform real-world actions across your software stack.

  1. Click Integrations in the sidebar.
  2. Select the service you want to connect (e.g., Stripe, Slack, Email/SMTP, CRM, GitHub).
  3. Follow the standard OAuth prompt or enter your API credentials.
  4. Once marked Connected, agents granted permission in their tool scope can securely interact with that service.

Workflow 5: Creating Custom Agents

Build purpose-built agents for unique company tasks.

  1. Go to Custom Agents and click + Create Agent.
  2. Basic Settings: Enter a name, description, and agent persona.
  3. Configure Capabilities:
  4. Skills: Attach relevant skills from your library (e.g., SEOAnalyzer, VendorVerifier).
  5. Tool Scope: Select allowable tools (e.g., Email, Database, Web Scraper).
  6. Trigger Keywords: Define keywords that activate this agent automatically.
  7. Behavior Prompt: Provide instructions, rules, and formatting expectations.
  8. Click Save & Deploy. You can test your agent interactively before rolling it out.

Workflow 6: Setting Policies & Guardrails

Enforce organizational safety rules to prevent unauthorized or unintended actions.

  1. Navigate to Policy & Guardrails.
  2. Click + New Policy Rule.
  3. Configure the rule parameters:
  4. Target Agents: Apply to all agents (*) or specific agents (e.g., FinanceAgent).
  5. Constraint Type: Choose from Maximum Transaction Amount, Rate Limit (actions/hr), Require Approval, or Time Window Restrictions.
  6. Approval Role: Designate which team role must sign off (e.g., Admin or Workspace Owner).
  7. Click Enable Policy to activate the rule immediately.

Workflow 7: Installing Marketplace Packages

  1. Go to Marketplace in the sidebar.
  2. Browse curated agent packages and community templates.
  3. Click on any package to view capabilities, ratings, and documentation.
  4. Click Install to add the agent package to your workspace.

Best Practices & Safety Tips

  • Start in Hybrid Mode: When launching new agents or integrations, use Hybrid mode to verify accuracy before enabling full autonomy.
  • Whitelist Domains: Always lock your customer-facing chat widgets to your verified website domains.
  • Keep Knowledge Updated: Refresh documents in the Memory Bank whenever policies, pricing, or product specs change.
  • Set Financial Thresholds: Use Policy Guardrails to mandate human approvals on transactions above defined limits.

Troubleshooting & Support

  • Agent not triggering an action? Check Policy & Guardrails to see if a restriction rule or missing integration is blocking execution.
  • Pending approvals accumulating? Adjust your policy sensitivity or delegate approval roles to additional team members.
  • Need help? Reach out to support@pluggedspace.org or visit pluggedspace.org.