Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.sigmic.ai/llms.txt

Use this file to discover all available pages before exploring further.

Console

The Console is a standalone web application that provides a visual interface for managing your Sigmic AI projects and configurations. It shares authentication with the main API, so you can use the same credentials. URL: https://console.sigmic.ai

Features

Organization & Team

Manage your organization, invite team members via email, assign roles, and control access across projects.

Project Management

Create, edit, enable/disable, and delete projects. Set a default project for your account.

Model Selection

Choose from Google Gemini, Anthropic Claude, Qwen, and DeepSeek models with a visual model picker.

MCP Server Config

Configure MCP servers with support for network (SSE/HTTP), stdio, and JSON input modes.

API Key Management

Create API keys with optional expiration, toggle enabled state, and manage per-project keys.

Skills Editor

Define custom agent skills with descriptions and instruction bodies.

Subagent Controls

Enable or disable built-in subagent capabilities per project.

Integrations

Connect external platforms like Slack. Map channels to projects with per-channel auto-execute and system prompts.

Email

Connect end-user email accounts via Google, Microsoft, or IMAP. Agents can read, search, send, and draft emails.

MCP Server Marketplace

Browse pre-configured MCP servers and install them into projects with secure credential management. Supports API key and OAuth authentication.

Skill Marketplace

Browse and install official skills into your projects with one click. Filter by category and manage installations across projects.

Knowledgebases

Create RAG knowledgebases, upload documents, and manage access tokens for MCP-based retrieval.

Getting Started

  1. Navigate to https://console.sigmic.ai
  2. Sign in with your existing Sigmic AI credentials (same as the Chat UI)
  3. You’ll see your project list on the dashboard

Organization & Team

The Organization page lets you manage your team and control who has access to your projects.

Organization Settings

  • Name — Rename your organization (admin only)
  • Slug — Your organization’s unique URL identifier

Inviting Members

Admins can invite new team members by email:
  1. Go to the Organization page
  2. Enter the email address and click Send Invite
  3. The invited person receives an email with a registration link
  4. Once they register, they automatically join your organization as a member
Pending invitations appear in the member list with a “Pending” badge and can be revoked at any time.

Managing Members

The member list shows all accepted members and pending invitations. Admins can:
  • Change roles — Promote a member to admin or demote an admin to member
  • Remove members — Remove a member from the organization (also removes all their project access)

Roles

RoleOrganizationProjects
AdminInvite/remove members, change roles, rename orgImplicit admin access to all projects
MemberView member listAccess only to explicitly assigned projects

Project Settings

Each project has a Settings page with the following tabs:

General

  • Model - Select from available AI models across multiple providers
  • Auto-execute (YOLO mode) - Toggle automatic tool approval
  • Default Tool Policy - Set the default approval policy for tools
  • Workspace Path - Configure the base workspace path
  • Daily Token Limit - Set per-project token usage limits

MCP Servers

Configure Model Context Protocol servers that provide tools to the agent. Supports three input modes:
  • Network - Connect to HTTP/SSE MCP servers by URL
  • Stdio - Run local command-based MCP servers
  • JSON - Paste raw JSON configuration

Skills

Define custom skills that extend the agent’s capabilities. Each skill has:
  • Name - Unique identifier
  • Description - What the skill does
  • Body - Detailed instructions for the agent

Prompts

Add custom system prompt text that is appended to the base prompt for all sessions in this project.

Subagents

Toggle built-in subagent capabilities on or off per project. Subagents are specialized sub-agents that the main agent can delegate tasks to.

API Keys

Create and manage API keys scoped to this project. Keys can have optional expiration dates and can be toggled on/off.
API key secrets are only shown once at creation time. Make sure to copy and store them securely.

Members

Control who can access this project. Only visible to project admins (and org admins who have implicit access to all projects).
  • View members — See all users with access to this project and their roles
  • Add members — Add existing organization members to the project (select from a dropdown of org members not yet on the project)
  • Change roles — Set a member as project admin or project member
  • Remove members — Revoke a member’s access to this project
Project RoleCapabilities
AdminManage project members, full configuration access
MemberUse the project for chat and tasks, view configuration
You can only add users who are already members of your organization. To add someone new, first invite them from the Organization page, then add them to specific projects.

Integrations

Connect external messaging platforms to Sigmic AI. Currently supports Slack with more platforms coming soon.
  • Slack - Install the Sigmic AI Slack app to your workspace via OAuth
  • Channel Mapping - Route specific Slack channels to specific projects
  • Per-Channel Config - Set auto-execute mode and custom system prompts per channel
  • Channel Discovery - Browse available Slack channels to set up mappings
  • Thread Follow-ups - Replies in Slack threads continue the existing conversation
To set up Slack integration:
  1. Go to the Integrations page in the Console
  2. Click Add Slack to initiate the OAuth installation
  3. Authorize the app in your Slack workspace
  4. Map channels to projects and configure per-channel settings

Knowledgebases

Create RAG (Retrieval-Augmented Generation) knowledgebases backed by Ragie. Upload documents and the agent can search them via MCP.
  • Document Upload - Upload PDFs, docs, and text files
  • Status Tracking - Monitor document processing status
  • Access Tokens - Create scoped access tokens for MCP server authentication
  • Auto-MCP - Knowledgebases are automatically available as MCP tools to the agent

Email

The Email page lets you connect end-user email accounts to Sigmic AI so agents can read, search, send, and draft emails on behalf of your users. Email access is managed through a two-level model: providers (configured once by you) and accounts (linked per end-user).

Email Providers

A provider represents an email service your organization uses. You create one provider per service, then link individual end-user accounts to it.
Provider TypeSetupUse Case
GoogleOAuth via Google WorkspaceGmail accounts
MicrosoftOAuth via Microsoft 365Outlook / Exchange Online
IMAPServer hostname and portOn-premise mail servers, other providers
To add a provider:
  1. Go to the Email page in the Console
  2. Click Add Provider
  3. Choose a provider type and enter a name
  4. For Google or Microsoft, complete the OAuth authorization flow
  5. For IMAP, enter the IMAP and SMTP server details

Email Accounts

An email account links one of your end-users to a provider. Each end-user can have one email account. When creating an account, you configure:
  • End-User ID — Your internal identifier for the user (the same ID you pass in endUserId when creating tasks or widget sessions)
  • Email Address — The email address the agent will access
  • Display Name — Optional friendly name
  • Scopes — Fine-grained permissions for what the agent can do:
ScopeAgent Capabilities
readList, search, and read emails. List folders. View threads. Download attachments.
sendSend new emails and reply to threads
draftCreate email drafts for the user to review before sending
You can update scopes or suspend an account at any time without deleting it.

Enabling Email on a Project

Email is opt-in per project. To enable it:
  1. On the Email page, click Add to Project
  2. Toggle email on for the desired projects
When email is enabled on a project and the end-user has an active email account, the agent automatically gains access to email tools during task execution — no additional configuration needed.

What the Agent Can Do

Once email is enabled, the agent has access to these capabilities (subject to the account’s scopes):
  • Search and read — Find emails by sender, recipient, subject, date range, or keywords. Read full email content including attachments.
  • Send and reply — Compose new emails or reply within existing threads, with CC handling.
  • Draft — Create drafts that the end-user can review and send manually.
  • Organize — List folders/labels, view full threads, and toggle read/unread status.
Email tools are automatically injected into agent sessions when the project has email enabled and the end-user has an active account. You don’t need to configure an MCP server manually.

MCP Server Marketplace

The MCP Server Marketplace provides a catalog of pre-configured MCP servers that you can install into your projects. Instead of manually configuring server URLs, commands, and credentials, the marketplace handles everything — you just provide your API key or complete an OAuth flow, and the server is ready to use. Navigate to MCP Servers from the Console sidebar.

Browsing Servers

The marketplace displays servers in a card grid. Each card shows the server’s name, description, category, tags, version, and authentication type. You can:
  • Filter by category — Developer Tools, Productivity, CRM, Ecommerce, Data, or Communication
  • Search — Find servers by name, description, or tags
  • View connection status — See which projects each server is connected to

Authentication Types

Each MCP server has one of three authentication types:
Auth TypeBadgeWhat You Do
API KeyYellowEnter your API key in a credential form
OAuthBlueComplete an OAuth authorization flow with the provider
NoneGrayInstall directly with no credentials needed

Installing a Server

  1. Click Connect on any server card
  2. A modal shows all your projects — click Install or Connect next to the project
  3. For API Key servers: fill in the credential form (e.g., paste your Brave Search API key)
  4. For OAuth servers: you’ll be redirected to the provider to authorize, then redirected back automatically
  5. For No Auth servers: the install happens immediately
The server is immediately available to agents in that project. Credentials are encrypted at rest and automatically injected into agent sessions.

Managing Installations

  • Servers already installed show a Connected to: Project A, Project B label on their card
  • Click Manage to open the project picker and install or remove the server from any project
  • Use Update Credentials to rotate API keys without uninstalling
  • Removing a server from a project does not affect other projects

Credential Security

All credentials are encrypted using AES-256-GCM before being stored. They are never exposed in API responses after initial submission. At session time, the system decrypts credentials and injects them into the server’s environment variables, headers, or URL as needed. For OAuth servers, tokens are automatically refreshed when they expire.

Available Servers

ServerCategoryAuthDescription
Brave SearchDeveloper ToolsAPI KeySearch the web using the Brave Search API
More servers are being added regularly. Each server includes a link to the provider’s documentation for obtaining credentials.

Skill Marketplace

The Skill Marketplace provides a catalog of official, ready-to-use skills that you can install into any of your projects with one click. Skills are curated prompt templates that give the agent specialized capabilities. Navigate to the Marketplace from the Console sidebar.

Browsing Skills

The marketplace displays skills in a card grid. Each card shows the skill’s name, description, category, tags, and version. You can:
  • Filter by category — Operations, Marketing, Analytics, or Customer
  • Search — Find skills by name, description, or tags
  • View details — Click a skill name to see the full prompt and any bundled files

Available Skills

SkillCategoryDescription
Product Listing OptimizerMarketingOptimizes product titles, bullet points, descriptions, and backend keywords
PPC Campaign StrategistMarketingPlans, analyzes, and optimizes pay-per-click advertising campaigns
Inventory PlannerOperationsForecasts demand, plans reorder quantities, and monitors stock levels
Pricing StrategistAnalyticsAnalyzes competitor pricing and develops repricing strategies
Customer Comms WriterCustomerDrafts customer service responses, follow-ups, and dispute resolution messages
Sales Analytics AdvisorAnalyticsAnalyzes sales data, identifies trends, and generates business intelligence reports
Order & Fulfillment ManagerOperationsManages order processing, shipping logistics, and returns handling
Financial AnalystAnalyticsProvides P&L statements, cash flow projections, and profitability analysis
Email & Social MarketerMarketingCreates email campaigns, social media content, and marketing strategies
Review & Feedback AnalystCustomerAnalyzes customer reviews to identify product improvement opportunities

Installing a Skill

  1. Click Install on any skill card
  2. A modal shows all your projects — click Install next to each project that should have this skill
  3. The skill is immediately available to agents in that project
Installed skills appear alongside any custom skills you’ve defined in the project’s Skills tab.

Managing Installations

  • Skills already installed show an Installed in: Project A, Project B label on their card
  • Click Manage to open the project picker and install or remove the skill from any project
  • Removing a skill from a project does not affect other projects

Bundled Files

Some skills include bundled template files (e.g., Dockerfile, CI configs, tsconfig.json). These files are available to the agent when the skill is active, allowing it to scaffold projects with real configuration files rather than generating them from scratch.

Version Tracking

Each skill has a version number. When a newer version is available for a skill you’ve installed, you can update it to get the latest prompt improvements and template files.