Agents

Configure specialized AI agents in DarcyIQ that combine custom instructions, MCP tool integrations, calibration forms, and automated scheduling. Build purpose-built agents for your team and let them run tasks on your behalf — on demand or on a recurring schedule.

From Chat to Automation: Create a "Sales Analyst" agent with product knowledge and CRM tools, then schedule it to generate pipeline reports every Monday morning.

Overview

Agents elevate DarcyIQ from a general-purpose AI into a library of specialized experts, each tailored to a specific role and equipped with the tools it needs.

Feature	Capability	Business Impact
Custom Instructions	Define the agent's role, behavior, and response style	Consistent, role-appropriate interactions
MCP Integrations	Attach MCP tools so the agent can interact with external systems	Agents that take action, not just answer questions
Calibration Forms	Present structured input forms to users before a conversation	Collect context up-front for better results
Agent Types	Chat agents for DarcyIQ, Widget agents for website embedding	Deploy AI where your users are
Automated Scheduling	Run agents on a recurring basis with full context	Hands-off reporting, monitoring, and task execution
Team Sharing	Share agents with configurable permissions	Standardized expertise across the organization

Creating an Agent

1

Open the Agents Panel Click the Agents icon in the sidebar to open the Agents drawer, then click New Agent.

2

Name Your Agent Give the agent a clear, descriptive name (e.g., "Technical Writer", "Sales Analyst", "AWS Advisor").

3

Choose an Agent Type

Type	Description	Use Case
Default	Regular chat agent inside DarcyIQ	Internal team use
Widget	Embeddable agent for your website	Customer-facing support or self-service

4

Write Role and Instructions Provide detailed instructions that tell the agent who it is, how it should behave, what tone to use, and what topics it should focus on. This is the core of the agent's personality and expertise.

5

Configure Optional Settings Add welcome screen options, calibration forms, MCP integrations, and schedules (all described below).

6

Save and Share Click Create Agent to save. You can then share the agent with your team.

Agent Configuration

Role and Instructions

The Role and Instructions field is the most important part of your agent. Write detailed guidelines covering:

• The agent's persona and expertise area

• Communication tone and style (formal, friendly, technical)

• What the agent should and should not do

• Preferred response formats (bullet points, tables, structured documents)

• Domain-specific knowledge or constraints

Pro Tip: Be specific in your instructions. "Always include 3 supporting data points" is better than "Be thorough." Instructions up to 10,000 characters are supported.

Welcome Screen Options

Optionally configure a use case label and initial prompt so the agent appears on the DarcyIQ welcome screen with a quick-start button. Both fields must be provided together.

Setting	Purpose	Example
Use Case	Short label shown on the welcome screen	"Technical Documentation"
Initial Prompt	Pre-filled message when the user clicks the agent	"Help me write API documentation for..."

Calibration Forms

Calibration forms present structured input fields to the user above the chat input when the agent is selected. Use them to collect key context before the conversation begins.

Configuration	Description
Form Title	Header displayed above the form fields
Form Icon	Icon displayed alongside the title
Fields	Custom fields the user fills in (text, dropdowns, etc.)

This is useful for agents that need specific information to do their job well — for example, a proposal writer that needs the client name, budget range, and timeline before drafting.

MCP Integrations

Attach MCP tools to give your agent access to external systems and data sources. When a user chats with the agent, the selected integrations are automatically available.

To configure:

1. Open the agent editor

2. Scroll to MCP Integrations

3. Check the integrations you want to enable

The agent will be able to use any tools provided by the selected MCP integrations during conversations and scheduled runs.

Managing Agents

Agents Drawer

The Agents drawer is the hub for agents and everything they use:

Tab	Purpose
Agents	View, create, edit, and manage your agents
To Do's & Tasks	View task submissions and their results
Automation	Manage all recurring schedules (see below)
Skills	Create and manage reusable skills for your agents
Integrations	Connect and manage the tools DarcyIQ can use

Sending an Agent a Task

Click any agent card on the Agents tab to open a task composer — Send a task to this agent. Describe what you need and submit; DarcyIQ confirms with Assigned to {agent} and switches to the To Do's & Tasks tab, where the run appears as it progresses. The agent works in the background, so you can close the drawer and carry on.

Agent List Features

• Grid or Table view: Switch between card and table layouts

• Search: Filter agents by name

• "Only mine" filter: Show only agents you created

• Default agent: Star an agent to set it as your default for new conversations

• Share: Invite team members with Owner, Editor, or Read access

• Edit: Modify agents you own

• Archive: Remove an agent from the active list, recoverable from the Archived view

Sharing & Permissions

Click the Share button on any agent you own to invite team members.

Permission	Capabilities
Owner	Full control — edit, share, delete
Editor	Modify the agent's configuration
Read	Run the agent without changing it

Recipients run a shared agent as themselves, using their own credentials and context rather than yours.

Shared Agents Bring Their Dependencies

An agent is only useful if the things it depends on come with it. When you share an agent, DarcyIQ also grants the recipient read-only access to that agent's integrations and custom skills, so it works for them the way it works for you.

The share dialog tells you exactly what's included before you confirm — for example 2 integrations, 1 skill included — grouped into Integrations and Skills so there are no surprises.

This is a real grant of access. Sharing an agent gives the recipient read-only visibility of the integrations and skills it uses, and that access can't be removed piece by piece while the agent is shared. If you don't want someone to reach a particular integration, don't share an agent that depends on it.

Included in a share	Not included
Integrations the agent has enabled	Knowledge bases
Custom skills the agent has been assigned	Lists

Knowledge bases and lists stay out because they're chosen per run rather than baked into the agent.

Revoking Access

Derived access follows the agent, so you manage it by managing the share:

• Un-share the agent and the recipient's access to its integrations and skills goes away with it.

• Archived agents grant nothing — archiving an agent withdraws the derived access until you restore it.

• On an integration's or skill's own sharing panel, people who reached it this way appear under Via shared agents, labeled Read-only via agent '{name}'. You can't remove them from there; un-share the agent instead.

Before You Remove a Dependency

Because agents depend on integrations and skills, removing one can quietly break agents — including other people's. DarcyIQ warns you before you confirm.

Archiving or deleting a skill, or disconnecting an integration, shows how many agents rely on it and names them, marking any that are shared. Recipients of a shared agent lose the capability silently, so the warning is worth reading closely.

Archiving & Deleting Agents

Deleting an agent from the active list archives it. Archived agents are hidden from the list, keep working state, and can be restored from the Archived view. Permanent deletion is a separate, deliberate action available only from that view.

Because agents power automations, archiving has knock-on effects that DarcyIQ handles for you:

Action	What happens to schedules and automations
Archive an agent	Its schedules are paused and any automations that launch it are deactivated. The dialog lists which automations will be affected before you confirm
Restore an agent	Schedules are re-enabled. Automations stay switched off so nothing unexpected fires — turn them back on when you're ready
Delete permanently	Automations referencing the agent stay disabled by default. Tick Also permanently delete this automation to remove them too

Restoring an agent deliberately does not re-enable its automations. Check the Automation tab after restoring and switch back on the ones you still want.

An automation whose agent has been archived shows a disabled toggle explaining the fix — unarchive the agent, or edit the automation and pick a different one.

Agents tied to a project (FDE agents) can't be archived on their own; archive the project instead.

---

Scheduling Agents

Schedule agents to run tasks automatically on a recurring basis — no manual intervention required. Scheduled runs execute in the background and can include meetings, project context, and integrations.

Creating a Schedule

Use the step-by-step schedule wizard to set up a new schedule:

1

Select an Agent Choose which agent will execute the scheduled task. Search by name to find the right one.

2

Set the Frequency Choose when the schedule should run using built-in presets:

Interval options:

Frequency	Description
Every 15 minutes	High-frequency monitoring or polling
Every 30 minutes	Regular check-ins
Every hour	Hourly updates
Every 2 hours	Moderate-frequency tasks
Every 6 hours	Periodic summaries
Every 12 hours	Twice-daily reports

Time-based options:

Frequency	Configuration
Daily	Pick a specific time of day
Weekly	Pick a day of the week and a time
Monthly	Pick a day of the month (1–28) and a time

All schedules run in your local timezone.

3

Define the Task Describe what the agent should do on each run. Write the task just like you would type a message in chat.

You can enrich the task with additional context:

Context	How to Add	Purpose
Meetings	Attach one or more meetings	Give the agent access to meeting transcripts and notes
Project	Link a project	Provide project-specific knowledge and documents
Integrations	Enable MCP tools, AWS, Atlassian, or Salesforce	Let the agent interact with external systems

Schedule Options

Option	Description
Enable / Disable	Toggle the schedule on or off without deleting it
Email on Completion	Receive an email when the scheduled run finishes. Add one or more recipient email addresses.

Managing Schedules

Open the Schedules tab in the Agents drawer to view and manage all schedules across your organization.

Feature	Description
Search	Find schedules by agent name or task description
Filter by type	Show only agent runs or workflow runs
Enable / Disable	Toggle individual schedules on or off
Edit	Modify the frequency and task details (the agent stays the same)
Delete	Permanently remove a schedule
Execution History	Click any schedule to expand and view recent executions with time, status (Success / Failed / Running), and duration

---

Use Cases

Sales & Revenue

• Pipeline Reporter: Schedule a daily summary of CRM pipeline changes

• Lead Researcher: Run competitive intelligence on new prospects every morning

• Meeting Prep: Automatically generate briefing docs before client calls

Operations & Monitoring

• System Health Check: Run infrastructure checks every 15 minutes

• Compliance Monitor: Weekly audit of policy adherence

• Cost Analyzer: Daily AWS spend summaries with anomaly detection

Customer Success

• Renewal Tracker: Monthly renewal risk reports

• Support Digest: Daily summary of open tickets and trends

• Onboarding Checklist: Automated status checks on new customer implementations

AWS Marketplace & Partner Programs

• Marketplace Manager: Update product listings, descriptions, and pricing through conversation

• Offer Coordinator: Create and track private offers for customers on a schedule

• Listing Auditor: Weekly analysis of marketplace listings to ensure completeness and optimization

Content & Documentation

• Technical Writer: Scheduled documentation reviews and updates

• Report Generator: Weekly market analysis or project status reports

• Knowledge Base Curator: Monthly review of outdated content

Integration with Other Features

Chat Integration

• Select any agent from the chat agent picker

• Each agent maintains its own conversation context

• Switch between agents without losing history

Project Integration

• Link agents to specific projects for contextual conversations

• Scheduled runs can reference project knowledge bases

Workflow Integration

• Agents can run as part of automated workflows

• Combine agent expertise with workflow automation for complex processes

Meeting Integration

• Include meeting transcripts as context for agent conversations

• Schedule post-meeting analysis to run automatically

Best Practices

Designing Effective Agents

1. Single Responsibility: Each agent should have a clear, focused role

2. Detailed Instructions: Provide specific guidelines, not vague directions

3. Right Tools: Only enable the MCP integrations the agent actually needs

4. Test First: Try the agent in chat before setting up schedules

5. Iterate: Refine instructions based on output quality

Scheduling Tips

Practice	Recommendation
Start conservative	Begin with daily schedules before moving to higher frequencies
Be specific	Write clear task descriptions — the agent performs exactly what you ask
Add context	Attach meetings and projects to give the agent the information it needs
Monitor results	Review execution history regularly to ensure quality
Use notifications	Enable email notifications so you know when runs complete

Pro Tip: Start with one well-defined agent for your most common use case. Schedule it to run a daily task, review the output quality, and then expand to more agents and higher frequencies.