Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.usesyndicate.org/llms.txt

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

Agents are specialized AI workers. Each agent has a name, specialization, model, instructions, tools, optional skills, and optional MCP access.

Agent anatomy

AreaWhat it controls
IdentityAgent name and specialization
SoulThe agent’s role, behavior, boundaries, and working style
ToolsWhich actions the agent can use
SkillsSpecialized workflows bundled into the agent’s instructions
MCP serversInstalled MCP servers explicitly enabled for that agent
ModelThe provider model used when the agent runs

Specializations

Syndicate includes built-in specializations such as:
  • Manager
  • Frontend Developer
  • Backend Developer
  • QA Engineer
  • DevOps Engineer
  • Technical Writer
  • Researcher
  • Finance Analyst
  • Data Scientist
  • Marketing Strategist
  • E-Commerce Specialist
  • AI/ML Engineer
  • Mobile Developer
  • Game Developer
Some areas split into more focused roles in the marketplace, such as UI Designer, Test Engineer, Site Reliability Engineer, AI Engineer, ML Engineer, and Security Engineer.

Soul

The Soul is the most important agent instruction file. It tells the agent:
  • Who it is.
  • What it is responsible for.
  • What quality bar it should hold.
  • What it should avoid.
  • How it should communicate.
Good Soul instructions are specific. Avoid generic traits like “be helpful” and prefer concrete rules such as “include file paths when reporting code changes.”

Tools

Tools determine what the agent can do. For example, a research-only agent might be allowed to read and search files but not edit them. A coding agent usually needs edit and command permissions.

Skills

Skills add domain-specific workflows. A Technical Writer might use documentation skills, while a QA Engineer might use test planning, browser testing, and regression-check skills.

MCP access

If MCP servers are installed in Settings, you can enable specific servers per agent. Keep access scoped to the agents that actually need it.