Getting startedCore conceptsWays to extend Syntic Code

Extend Syntic Code

Out of the box, Syntic Code reads and edits code, runs commands, looks things up online and hands work to subagents. These are the ways to give it more.

AddSo thatLives inPage
SkillsThe agent knows how your team does a task, and you can run it as /name.syntic/skills/<name>/SKILL.mdSkills & custom commands
Custom commandsYou can fire a saved prompt with arguments.syntic/commands/<name>.mdSkills & custom commands
HooksYour script runs every time, at a fixed point (format after edits, block risky commands)hooks in settingsHooks
MCP serversThe agent can use outside systems: databases, trackers, docs, browsers.mcp.jsonMCP servers
Custom agentsA specialist with its own instructions and tools.syntic/agents/<name>.mdSubagents
Output stylesReplies are written in a voice that suits yououtput-styles/*.mdOutput styles
PluginsAll of the above, packaged and sharedA marketplacePlugins
ModsA small program draws extra interface (footer, band, pane)~/.syntic/mods/Mods
MemoryProject knowledge loads into every sessionSYNTIC.mdMemory

Which one do I need?

  • “Every time it edits a file, run the formatter.” → a hook. It does not depend on the agent remembering.
  • “When we write a migration, follow these seven steps.” → a skill. The agent picks it up when relevant.
  • “Give me /standup that summarizes yesterday’s commits.” → a custom command.
  • “Let it read our Jira tickets.” → an MCP server.
  • “Always use pnpm, never npm.” → a line in SYNTIC.md.
  • “Share our setup with ten repositories.” → a plugin.