Practical overview
When it fits
Use auto-co when you want a persistent, terminal-driven Claude Code workflow for software development and delivery, and can provide human oversight for sensitive decisions and credentials.
Who it is for
- Developers who want a recurring Claude Code process to work through software tasks.
- Small teams or project builders interested in coordinating role-based agents across planning, implementation, QA and deployment.
Documented uses
- Coordinate agent discussions and decisions around a software project's next steps.
- Continue development work across repeated cycles using persistent shared Markdown state.
- Monitor cycle output, status, health, history and exported project data.
Requirements & limits
The repository is public and MIT-licensed. Claude Code installed and working is the stated prerequisite. An Anthropic API key is needed if Claude Code's built-in authentication is not used.
- The workflow depends on Claude Code rather than providing a model-independent agent layer.
- Shared state is kept in Markdown files, with structured logs in JSONL; the project says it uses no database or server.
- The README describes a 14-agent roster, selecting 3–5 relevant agents per cycle.
- Optional settings include loop interval, idle detection, Telegram notifications and webhooks.
- Human input may be needed for credentials, spending decisions or legal questions; the project says spending requires approval.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Set up an auto-co project
Create a project from the published template and start its recurring Claude Code workflow.
Before you start
- Claude Code installed and working.
- Use Claude Code built-in authentication, or configure an Anthropic API key if that authentication is not being used.
- A terminal with npx available, as shown in the documented setup commands.
Steps
Run the command below to create a project.
npx create-auto-co my-companyChange into the project directory with
cd my-company.Start the workflow with the command below.
make startUse the command below to view cycle output.
make monitor
Expected outcome
A project is created and the recurring workflow is started, with cycle output available through the monitor command.
Guide limitations
- The instructions describe setup and monitoring; they do not guarantee a particular project outcome or deployment.
- The README describes ongoing human input for credentials, spending decisions and legal questions.
Compare options for Agent orchestration
These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.
| Project | Documented purpose | Platform | Access and requirements |
|---|---|---|---|
| auto-co | Coordinate agent discussions and decisions around a software project's next steps. | CLI | The repository is public and MIT-licensed. Claude Code installed and working is the stated prerequisite. An Anthropic API key is needed if Claude Code's built-in authentication is not used. |
| Agentic AI Helpdesk Platform | Classify tickets by issue domain and category, and estimate urgency and complexity. | Cloud, Docker, Web | |
| Tgo | Configure multiple AI agents for different customer-service scenarios and connect supported LLM providers. | Docker, Self-hosted, Server | The repository describes Tgo as open source and provides self-hosting instructions. Its README lists minimum server requirements of four CPU cores and 8 GiB of RAM; local development setup lists macOS, Linux, and WSL2. |
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 7 October 2026
- Original website title
- NikitaDmitrieff/auto-co-meta · Official homepage title · retained in its source language