Skip to content

[RSI Baseline] Provide a persistent FlyBot instance for integration experiments #785

Open

Description

@zy95-12

Parent Epic: #784

Why

The RSI work needs a real, persistent FlyBot instance to validate deployment, runtime behavior, task recovery and dashboard integration. A local mock alone cannot establish that the full system works in a running environment.

Request to maintainers

Please provide access to a running FlyBot instance for authorized RSI integration experiments. Specific port(s), access address and connection details can be coordinated privately over WeChat; do not post credentials or sensitive endpoints in this public issue.

Scope

  • Identify the instance owner, FlyBot commit/version, enabled plugins and deployment mode.
  • Provide an authenticated, least-privilege access path and document permitted actions.
  • Confirm persistence locations, restart behavior, health checks/logs and how to pause experiments.
  • Define a safe test workspace/project and data-retention expectations.
  • Document access for the cloud Runner / Code Agent, if included in the test setup.

Acceptance criteria

  • The RSI team can connect through the agreed private channel.
  • Access is authenticated, scoped and revocable.
  • Version/configuration and permitted operations are documented.
  • A harmless smoke test and restart/recovery check can be performed without touching production data.
  • No credentials, tokens or private connection details are committed to the repository.

This is an environment/access prerequisite, not a request to expose an unauthenticated port to the public internet.

Activity

  1. added
    area:automationScheduled tasks and automation
    enhancementNew feature or request
    needs-humanThe agent loop stopped; a person must decide (reason in the last comment)
    on Oct 22, 2026
  2. Birfy commented on Oct 22, 2026

    @Birfy
    Collaborator

    Written-by: Triage

    enhancement|area:automation|needs-human. Blocking question is not a design one, it is an owner grant: who hosts the instance and who may use it. Nothing in this repository can answer that, and the issue itself says the details are to be agreed privately — which is fine, but it means this cannot be an agent-ready code change.

    Two facts that move it forward: the deployment already exists and is documented (Dockerfile, compose.yaml, docs/docker.md, and the release tarballs from v0.2.4), so "install one more instance" is not research; and the second instance for the self-test already runs on fixed homes and ports (@selftesta2.home / @selftestb2.home, 3096/3097), so the pattern to copy is written down in docs/self-test.md.

    To become agent-ready it needs: the owner's grant (who, which host, which version), one acceptance sentence that a non-author can check (e.g. "a fresh container from the v0.2.4 tarball answers a health call and survives a restart with its state"), and confirmation that nothing secret lands in this repository. The rest is documentation that can ship.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    area:automationScheduled tasks and automationenhancementNew feature or requestneeds-humanThe agent loop stopped; a person must decide (reason in the last comment)

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions