Skip to main content
Agents are Markdown files with YAML frontmatter that define a system prompt, a set of allowed tools, and an operating mode. The cyberstrike agent command helps you generate new agents using an LLM and list every agent available in your current project and global configuration. 
cyberstrike agent <subcommand>

​
Subcommands

SubcommandDescription
cyberstrike agent createGenerate a new agent from a description
cyberstrike agent listList all available agents

​
cyberstrike agent create

Generates a new agent file. When all required flags are provided the command runs non-interactively; otherwise it prompts for missing values. 
cyberstrike agent create [flags]

​
Flags

​
--description
string
What the agent should do. Passed to the model to generate the system prompt and identifier. Required when running non-interactively.
​
--path
string
Directory path in which to write the agent file. When omitted, the command prompts you to choose between the current project (.cyberstrike/agent/) and the global config directory.
​
--mode
string
Agent operating mode. Choices: all, primary, subagent.
  • all — can function as both a primary agent and a subagent.
  • primary — acts only as the top-level agent driving a session.
  • subagent — can be dispatched by other agents via the task tool.
​
--tools
string
Comma-separated list of tools to enable. Pass an empty string to disable all tools. When omitted, all tools are enabled.Available tools: bash, read, write, edit, list, glob, grep, webfetch, task, todowrite, todoread, report_vulnerability
​
--model, -m
string
Model to use for generation, in provider/model format. Defaults to the globally configured model.

​
Examples

cyberstrike agent create

​
Output

When running non-interactively, the command prints the path of the created file to stdout and exits. 
/home/user/.config/cyberstrike/agent/rest-api-tester.md

​
cyberstrike agent list

Lists every agent found in the current project and global configuration directories, sorted with built-in (native) agents first, then alphabetically. 
cyberstrike agent list

​
Output

Each entry shows the agent name, its mode, and its full permission ruleset as a JSON array. 
web-application (subagent)
  [{"permission":"bash","action":"allow","pattern":"*"},...]
cyberstrike (primary)
  [{"permission":"*","action":"allow","pattern":"*"},...]
rest-api-tester (primary)
  []
Use the agent name with cyberstrike run --agent <name> or select it in the TUI to start a session with that agent.

Build docs developers (and LLMs) love

Get started for freeTalk to us