INSTALLATION GUIDE · VERIFIED 2026-08-05

Install MonkeyCode without skipping the production decisions.

The project publishes a one-line online installer and minimum infrastructure figures. Use them as an evaluation starting point—not as a production capacity or security guarantee.

DIRECT ANSWERMonkeyCode can be installed on infrastructure operated by your team.

The current project README documents separate console and development-host capacity and provides this online installation command:

bash -c "$(curl -fsSL 'https://monkeycode-ai.com/online/install')"

Review any remote script before executing it, test in an isolated environment, and verify the current command in the project README. Installation commands can change.

Documented starting point

Two infrastructure roles, not one generic server.

These are published minimums. Concurrent environments, repository size, builds, previews, models, logs, and retention can increase the requirement.

MonkeyCode console2 cores

4 GB memory · 40 GB storage

Development environment host8 cores

16 GB memory · 100 GB storage

Installation sequence

Move from a disposable test to an accepted deployment.

The installer starts the process. Your acceptance checks determine whether the result is suitable for team use.

  1. STEP 1
    Prepare isolated infrastructure

    Allocate the console and development host, define DNS and TLS, and document allowed inbound and outbound routes.

  2. STEP 2
    Review and run the current installer

    Inspect the downloaded script, confirm its source and checksum where available, then execute it in a test environment.

  3. STEP 3
    Configure Git and model access

    Use scoped credentials, decide where model requests go, and verify that secrets do not appear in task output or logs.

  4. STEP 4
    Run an acceptance task

    Clone a non-sensitive repository, build it, run tests, inspect the diff, and confirm that another reviewer can reproduce the result.

  5. STEP 5
    Plan operations before rollout

    Test backup, restore, upgrades, rollback, host replacement, capacity alerts, retention, and incident ownership.

Production gate

Do not infer these controls from “self-hosted.”

Verify each control in your actual release and topology.

Network boundaries

Model APIs, Git hosts, package registries, previews, updates, and telemetry.

Identity and secrets

Administrator access, repository credentials, model keys, rotation, and revocation.

Isolation

Task-to-task, project-to-project, host, filesystem, process, and network isolation.

Operations

Logs, retention, backups, restore tests, upgrades, rollback, and vulnerability response.

NEXT DECISION

Validate security and capacity before adding production code.