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

Evo

Evo is a Python-based command-line developer tool that converts an existing codebase into an autoresearch loop. It discovers what to measure, instruments a benchmark and uses tree search with parallel subagents to investigate code changes and optimization opportunities.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Evo — 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 Evo when a repository has a measurable target and candidate changes can be evaluated repeatedly by a benchmark or gate. Its documented example is optimizing a JSON parser for speed; the same workflow can be adapted to other project-defined metrics where the evaluation is available.

Who it is for

  • Developers optimizing an existing codebase against a measurable performance or quality target.
  • Teams using supported coding-agent hosts and needing parallel repository experiments.
  • Researchers or engineers exploring repository-specific autoresearch workflows.

Documented uses

  • Discover a benchmark or evaluation target in a repository.
  • Instrument experiments for a measurable code optimization goal.
  • Explore multiple code-change branches with parallel subagents.
  • Reject changes that fail configured tests, invariants or score gates.
  • Monitor experiment progress through the dashboard.
Before you choose

Requirements & limits

Evo is presented as an installable CLI and self-hosted developer tool. The public repository identifies Python as its primary language and displays an Apache-2.0 license.

  • A repeatable benchmark or evaluation is central to the workflow.
  • Gates are important when a higher score could otherwise trade away correctness.
  • Host choice affects invocation syntax and installation steps.
  • Local and remote execution are both described, but remote providers require matching extras.
  • Parallel experiments use isolated Git worktrees and shared state.

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

Set up Evo for a repository optimization loop

Install the Evo CLI, connect a supported coding-agent host, discover a benchmark and start the documented optimization loop.

Before you start

  • A codebase to optimize.
  • A supported host such as Claude Code, Codex, Cursor, Hermes, Kimi, Opencode, OpenClaw or Pi.
  • The relevant host CLI, if it is not already installed.
  • For remote execution, the matching Evo provider extra.

Steps

  1. Install the CLI with the command below.

    Documented command
    uv tool install evo-hq-cli

  2. Install or select a supported host CLI, then run evo install to add the plugin and host hooks.

  3. Run /evo:discover in the host, optionally specifying a target such as the command below.

    Documented command
    make the JSON parser at src/parser.py faster

  4. Run /evo:optimize to start the experiment loop.

Expected outcome

Evo discovers or configures the evaluation and launches benchmarked optimization experiments through its agent loop.

Guide limitations

  • The invocation syntax varies by host.
  • The documentation describes benchmark instrumentation and execution but does not specify a general benchmark integration interface.
  • Remote backends require the corresponding provider extra.
Documentation 1 ↗
Evidence and transparency

Sources & review

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