forked from smallcloudai/refact
-
-
Notifications
You must be signed in to change notification settings - Fork 4
Integrations Command Line Service
refact-planner edited this page Jun 7, 2026
·
1 revision
Manage long-running local processes from Refact Agent.
Command-line service integrations wrap long-running processes such as dev servers, file watchers, tunnels, or log streams. The agent can start, restart, stop, or check the status of the service.
A service_* integration defines:
-
Command: the process to run. Use
%param_name%placeholders for model-filled values. - Working directory: optional service workdir. If empty, Refact uses the workspace directory.
- Description: what the model sees when deciding whether to call the service.
- Parameters: values the model can provide when starting the service.
- Startup wait port: a TCP port that indicates the service is ready.
- Startup wait keyword: text in stdout or stderr that indicates readiness.
- Startup wait timeout: maximum time to wait during startup.
- Output filter: limits and extracts useful logs from long service output.
The tool accepts start, restart, stop, and status. Status also returns accumulated stdout and stderr since the last check.
Use confirmation rules for services that modify data, open network listeners, or run expensive jobs. Prefer one-shot command-line tools for commands that finish immediately.
Refact on GitHub: https://github.com/JegernOUTT/refact
- Agent Modes
- Agent Tools
- Task Planner & Cards
- Worktrees
- Subagents
- Memory & Knowledge
- Hidden Roles & Plans
- Context Compression
- Scheduler & Cron
- Processes & PTY
- Buddy
- MCP
- Skills, Commands & Hooks
- Marketplace
- Chat System
- Providers
- Caps & Models
- Code Completion (FIM)
- AST
- VecDB
- Exec Runtime
- HTTP API
- Checkpoints & Git
- Voice