Appendix

Related Videos

• Demo video: View
• Launch livestream replay: View

Flocks Whitepaper

• Online preview: the embedded PDF below is the complete document
• Direct download: Flocks Whitepaper.pdf

Flocks Quick Installation Guide

• Online preview: the embedded PDF below is the complete document
• Direct download: Flocks Quick Installation Guide v2.0.pdf

If your browser cannot embed PDF previews, such as in some mobile or enterprise environments, use the "Direct download" link above.

CLI Reference

Run flocks --help to view the current CLI main commands. The most important commands for daily use are:

Command	Purpose	Common Scenario
flocks start	Start the backend and WebUI	First startup, daily operation
flocks stop	Stop the backend and WebUI	Before upgrade, troubleshooting, or restart
flocks restart	Restart the service	Restart after configuration changes or a required full restart
flocks status	View current runtime status	Confirm whether the service actually started
flocks logs	View backend and WebUI logs	Troubleshoot startup, model, tool, and page issues
flocks update	Upgrade to the latest version	CLI entry outside page upgrade or manual upgrade
flocks tui	Start the terminal interactive interface	Browserless environments, centralized terminal operation
flocks task	Manage Task Center capabilities	Queue tasks and scheduled tasks
flocks session	Manage sessions	Export, troubleshoot, or batch process sessions
flocks mcp	Manage MCP services	MCP integration and status checks
flocks skills	Manage Skills	Install, enable, and view Skills
flocks admin	Manage admin capabilities	User and administrator operations
flocks browser	Use the built-in browser runtime	Troubleshoot CDP browser automation
flocks export / flocks import	Export or import session data	Migration and backup
flocks stats	View usage statistics	Quickly confirm runtime data

The most common command combination is usually:

flocks start
flocks status
flocks logs

For upgrade or environment recovery, the common combination is:

flocks stop
flocks restart

Configuration Item Index

For common configuration domains in flocks.json or sample configuration, understand them by responsibility first instead of memorizing every field.

Configuration Domain	Purpose	Common Modification Scenario
provider	Defines model providers, adapters, and model sets	Connect OpenAI Compatible services, local models, or self-hosted gateways
api_services	Defines whether external security services are enabled	Connect TDP, ThreatBook, Qingteng, and other platforms
mcp	Defines MCP service integration	Extend external context or tool capabilities through MCP
channels	Defines messaging channels	Connect Feishu, WeCom, Slack, DingTalk, Telegram, WhatsApp, and email integrations
sandbox	Defines execution isolation mode	Team operation or stronger boundary control
server	Defines service-layer behavior	Adjust CORS or service access configuration
allowReadPaths	Defines allowed read paths	Explicitly authorize additional read paths
updater	Defines upgrade source and upgrade policy	Adjust GitHub / Gitee update sources

For first-time Flocks configuration, focus on provider, channels, and api_services. The remaining configuration is more relevant during extended integration, remote deployment, or team operation.

API Reference

The Flocks backend provides API services based on FastAPI. The complete API definition is the runtime OpenAPI definition:

• Swagger UI: http://127.0.0.1:5173/docs
• ReDoc: http://127.0.0.1:5173/redoc
• OpenAPI JSON: http://127.0.0.1:5173/openapi.json

For ordinary users, the most frequently used APIs usually focus on:

• Conversation management
• Model management
• Tools and MCP
• Workflow execution
• Task center
• Channel status and configuration
• Workspace file management
• Auth / Admin / Notifications
• Hooks, Hub, and Update

If you need complete API documentation, generate it from the current version's OpenAPI output to avoid version drift caused by manually maintained API lists.

Glossary

Rex

Flocks' default main Agent. It usually handles overall control, analysis, and scheduling.

Agent

An intelligent agent that can independently receive goals and execute tasks. It can complete work directly or delegate to sub-agents.

Workflow

A runtime unit that organizes a set of analysis steps, node relationships, and input/output constraints into a reusable process. It can be maintained manually or generated from natural-language descriptions.

Skill

A capability package that provides Agents with domain-specific knowledge, fixed processes, or integration specifications.

MCP

Model Context Protocol. It connects external context services or tool capabilities to Flocks through a unified protocol.

Workspace

The workspace Flocks uses at runtime to store inputs, outputs, intermediate artifacts, and reports.

Provider

A model provider or model adapter entry point. It describes where models come from and which interface is used to access them.

Task

A task unit queued, executed, and tracked by the system. It can appear as an immediate task, scheduled task, or API service task.

Channel

A channel capability that sends messages to external touchpoints such as Feishu, WeCom, Slack, DingTalk, Telegram, WhatsApp, and email.

Sandbox

A runtime isolation mechanism used to limit execution environments, permission scope, and resource boundaries.