OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Coding Agents

Bello

Bello is a Python-based, MIT-licensed CLI developer tool that coordinates coding agents with optional runtime supervision, completion review, adversarial testing and local log distillation. It can combine Codex or Claude Code with OpenAI, Anthropic and OpenRouter-backed components, assign different models to different roles, and run selected subagent work in parallel.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Bello — OpenViST project profile in Coding Agents.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Bello when a coding task benefits from configurable supervision and independent review around an existing coding agent. It is also relevant when different roles should use different model providers or when local log distillation is useful.

Who it is for

  • Developers using Codex or Claude Code for repository tasks
  • Teams designing multi-agent coding workflows
  • Builders who need separate implementation, review and adversarial-checking roles
  • Developers who want local reduction of noisy coding-agent tool output

Documented uses

  • Coordinate a coding task across a coder and optional supporting agents
  • Supervise live agent runs and redirect work that drifts from the task
  • Review completed work against its task and evidence
  • Probe implementations with edge cases, invalid input and feature interactions
  • Reduce tool-output noise locally before it reaches coding agents and subagents
Before you choose

Requirements & limits

Bello is presented as an open-source MIT-licensed Python CLI for self-hosted use. The documented setup requires Python 3.11 or newer and git, with additional operating-system and model-source requirements.

  • Choose which roles to enable independently: supervisor, reviewer, adversary and log distiller.
  • Different roles can use separate models and providers, which supports model-routing designs.
  • The log distiller runs locally, but its first enabled run downloads a 599 MB model.
  • Codex and Claude Code use subscription-based login flows in the documented setup; API-provider runtimes require Node.js 22.19 or newer.
  • Bello uses a disposable workspace for the coder and returns the final patch to the project.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install Bello and configure a model source

Prepare Bello for a coding-agent workflow.

Before you start

  • Python 3.11 or newer
  • git
  • macOS, Linux with bubblewrap, or supported native 64-bit Windows
  • A Codex subscription, Claude Code subscription, or API provider account
  • Node.js 22.19 or newer for API-provider runtimes

Steps

  1. Install Bello with pipx using pipx install bello.

  2. Run bello doctor to see what is ready.

  3. Configure one model source: log in through Codex, Claude Code, or an API-provider runtime.

  4. Install the optional log-distiller extra if local tool-output reduction is needed.

  5. Install the Codex or Claude Code plugin when Bello will be run from inside a coding agent.

Expected outcome

Bello and the selected model source are prepared for a coding task.

Guide limitations

  • The documented installation requirements differ by model source and operating system.
  • The log distiller downloads a 599 MB model on its first enabled run.
  • This guide covers setup rather than verification of a completed task.
Documentation 1 ↗
Source-based practical guide

Run a task through Bello from a coding agent

Use a coding agent to recommend and launch a Bello configuration for a repository task.

Before you start

  • Bello installed and a model source configured
  • The Codex or Claude Code plugin installed
  • A project containing task.md
  • A coding agent opened in that project

Steps

  1. Ask the coding agent whether it can see the Bello plugin.

  2. Ask it to recommend a price-and-quality balance for task.md.

  3. Review the resolved configuration it presents.

  4. Ask the agent to run task.md with the selected configuration.

  5. Review .supervisor/FINAL_REPORT.md for the result, changed files, checks and remaining risks.

Expected outcome

Bello runs the task through the selected coder and enabled supporting roles, then produces a final report.

Guide limitations

  • The selected roles, models and review rounds depend on the configuration.
  • The documented workflow returns a patch to the project after work in a disposable workspace.
  • The report records checks and remaining risks; it is not a guarantee that the implementation is correct.
Documentation 1 ↗
COMPARE OPTIONS

Compare options for AI coding assistance

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.

ProjectDocumented purposePlatformAccess and requirements
BelloCoordinate a coding task across a coder and optional supporting agentsCLI, Self-hostedBello is presented as an open-source MIT-licensed Python CLI for self-hosted use. The documented setup requires Python 3.11 or newer and git, with additional operating-system and model-source requirements.
SpectraClarify requirements and compare approaches before proposing a change.CLI, DesktopThe README lists a macOS and Windows desktop app, and a Homebrew installation for macOS. The CLI is published through npm and requires Node.js 22.14 or newer.
Skills — michaelshimelesInstall reusable task-specific skills into a coding-agent setup.CLIThe repository README provides an npx installation command for most coding agents and describes automatic or explicit skill invocation in Claude Code.
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
4 October 2026
Original website title
Makson179/Bello · Official homepage title · retained in its source language
How we review sourcesSuggest a correction