Manager.io CLI & AI Agent Skills — Open-Source Command-Line Tool for the API

1

I’ve put together an open-source CLI tool for Manager.io self-hosted edition that provides command-line access to the REST API v2. It also includes skill definitions for AI coding agents (Claude Code, OpenClaw, etc.).

It’s a single Bash script with no dependencies beyond curl and jq. Works on macOS, Linux, and WSL.

:link: GitHub: https://github.com/mprokopov/manager-ai-skills

This might be useful for these cases

  • Automation: script repetitive accounting tasks — bulk imports, periodic reports, data extraction — without clicking through the UI
  • Integration: connect Manager.io to other systems (CRMs, e-commerce platforms, custom dashboards) using simple shell commands
  • AI assistants: let an AI agent query your accounting data in natural language — useful for quick lookups without opening the browser
  • Remote monitoring: check balances, overdue invoices, or run reports from a terminal or over SSH
  • Backups & auditing: periodically export data in JSON for archival or external analysis

This is still a work in progress. Feedback, suggestions, and contributions are welcome.

Screenshot 2026-03-17 at 17.16.11
Screenshot 2026-03-17 at 17.16.111800×862 109 KB

5 Likes
2

Great effort @mprokopov

1 Like
3

Good to see some work with AI integration happening here.

I’m currently working on an MCP server, and one of the big problems we have is that the data attributes for the parameters is not well defined or even documented. My testing has also indicated that there’s also not a lot of input validation going on, as my test scripts that have been probing the API have been causing server crashes all over the place.

To help this work progress, I’ve put together an enhanced OpenAPI swagger file that includes a lot of the parameters in the schema. It’s not complete and may be incorrect in places, so I would welcome any comments or contributions.

The project is at https://github.com/SpiffyStores/manager-io-api-schema

2 Likes
4

@mprokopov very nice, appreciated :+1:

5

@mprokopov clean and clear code and good initiative. I decided to answer because it may be useful to highlight that if you use a cloud-based AI (Claude, GPT, Gemini, etc.), the AI will see the output of manager commands in your chat.

  • That means sensitive accounting data (invoices, customer balances, bank info, etc.) temporarily enters the AI provider’s servers as part of the conversation.

  • This is not the repo stealing data — it’s the inherent nature of using any cloud LLM with tools/shell access.

Mitigations:

  • Use the CLI standalone (no AI).

  • Or pair it with a local LLM that supports tool calling/shell execution (e.g., via Ollama + some agent framework) — data never leaves your machine.

6

Absolutely makes sense, AI and data security is the one big pain point at the moment

7

This definitely helps, thanks for sharing!