Skip to content

codeSplitters: hosted deploy follow-ups #248

Closed

Description

@QuinnsCode

Loose ends from #244:

  • Alchemy state: every deploy, desktop or hosted, clones into a fresh directory, so Alchemy starts with empty state. A stack without adopt may try to create resources that already exist. Keep state somewhere durable per repo and stage (R2, or Alchemy's own remote state).
  • Key form: while a deploy is running, the Deploy panel redraws every second and would clear a token half-typed into the key form. Leave the form alone while it's being edited.
  • Live output: a hosted step's output only arrives when the step finishes (one /exec per command), not as it streams. Stream it, or poll it from the container.
  • Key rotation: there's no way to re-seal stored keys under a new DEPLOY_SECRETS_KEY. Rotating it now means every owner has to set their key again.
  • More providers: only Cloudflare tokens can be stored. Railway or Hetzner box targets would need their own scoped tokens.

🤖 Generated with Claude Code

Activity

  1. QuinnsCode commented on Oct 10, 2026

    @QuinnsCode
    OwnerAuthor

    The Alchemy state item is done in the PR for #247. Hosted deploys keep each repo's state sealed in its DeployRunner, and desktop deploys keep it in ~/.codesplitters/alchemy//. Verified for real: a second hosted deploy from a fresh container updated the same Worker. The other items here are still open.

    🤖 Generated with Claude Code

  2. self-assigned this
    on Oct 10, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Labels

No labels
No labels

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions