Coding agents | Versalist Documentation

Select an agent guide. Each guide gives you install, authentication, key, configuration, verification, and revoke procedures.

Five hosts can run the Versalist Model Context Protocol (MCP) server. Pi uses command-line interface (CLI) commands only.

Agent guides

Use one guide for your current agent.

MCP hosts

OpenCode, Claude Code, Codex, Cursor, and Zed can start the MCP server.

Shared workflow

1 CLI

All six agents can run the same Versalist commands.

Select an agent

OpenCode

MCP and CLI

Run the Versalist Model Context Protocol server from an OpenCode project.

Launch with opencode

Project configuration: opencode.json

Set up OpenCode

Claude Code

MCP and CLI

Run the Versalist Model Context Protocol server from a Claude Code project.

Launch with claude

Project configuration: .mcp.json

Set up Claude Code

Codex

MCP and CLI

Run the Versalist Model Context Protocol server from a Codex command-line project.

Launch with codex

Project configuration: .codex/config.toml

Set up Codex

Cursor

MCP and CLI

Run the Versalist Model Context Protocol server from Cursor Agent.

Launch with agent

Project configuration: .cursor/mcp.json

Set up Cursor

CLI only

Use Versalist command-line interface commands from a Pi session.

Launch with pi

Project configuration: .pi/settings.json

Set up Pi

Zed

MCP and CLI

Run the Versalist Model Context Protocol server from the Zed Agent Panel.

Launch with zed .

Project configuration: .zed/settings.json

Set up Zed

Setup flow

Install the coding agent

Use the official installation command for your operating system.

Authenticate the coding agent

Use the agent account or an approved model provider credential.

Create a Versalist key

Create a key with only the scopes that your workflow requires.

Read the key guide

Connect Versalist

Add the MCP configuration, or use CLI commands if the host has no native MCP support.

Verify the setup

Use read-only status and list commands before you run a challenge.

Record an evaluation

Record a baseline, evaluate a candidate, and compare the results.

Shared CLI workflow

Install the Versalist CLI once. Each agent can then run the same repository commands.

Install the Versalist CLI

shell

npm install -g @versalist/cli
versalist --help

Record and compare runs

Replace the example commands, run identifiers, score values, and project web address.

shell

versalist list --search "agentic code optimization"
versalist start agentic-code-optimization-review

versalist run --command "python agent.py" --label baseline
versalist evaluate --run <baseline-run-id> --command "pytest" --score 70

versalist run --command "python agent.py" --label candidate
versalist evaluate --run <candidate-run-id> --command "pytest" --score 84

versalist compare --baseline <baseline-run-id> --candidate <candidate-run-id>
versalist submit --url https://github.com/you/solution --title "Candidate run"

The CLI stores local records under .versalist/. It does not upload these records automatically.

The submit command sends project metadata and URLs. It does not upload the contents of .versalist/.

Keep keys safe

Keep every key out of source control

Store VERSALIST_API_KEY in an environment variable or secret manager. Use a host environment reference when the host supports it.

Read key procedures

Create separate keys for separate tools or environments.
Give each key only the required scopes.
Do not put a key in a repository file.
Revoke a key when it is exposed, unused, or no longer required.

Start setup

Open an agent guide first. The guide links to account creation and scoped key setup.

For all CLI commands, open the Versalist CLI reference.