Skip to content

[skills] Add offload-to-sandbox skill - #326

Open
cramforce wants to merge 2 commits into
mainfrom
malte/offload-to-sandbox-skill
Open

cramforce wants to merge 2 commits into
mainfrom
malte/offload-to-sandbox-skill

Conversation

@cramforce

@cramforce cramforce commented Oct 10, 2026 •

Copy link
Copy Markdown
Contributor

Adds a new skill, Offloading work to a Vercel Sandbox (skills/offload-to-sandbox). It's about using the vercel sandbox CLI as remote workers for heavy builds, full test suites and many parallel (agent) tasks, so they don't load the local machine. It complements vercel-sandbox, which covers running untrusted code through the SDK.

The content comes from a multi-week performance program in which a dozen coding agents shared one laptop and offloaded builds and sharded test suites to Vercel Sandboxes. It is written to be project-agnostic:

  • when to offload and what to keep local
  • a base snapshot built once (pinned toolchain, pre-built baseline tree, helper scripts)
  • one sandbox per task: naming, sizing (2 GB per vCPU and the OOM risk), lifetime vs exec timeouts, tags
  • pushing source as a tarball of git-known files while keeping node_modules and build artifacts
  • running commands: quoting pitfalls (zsh globs and modifiers), script files, detached long jobs with an exit-status file
  • getting results back as small files
  • sharding suites; separating pre-existing failures from new ones against an untouched baseline
  • auth, a small wrapper script plus a SANDBOX.md for agents, cost and cleanup, and designing for restarts

Also adds the required ⤳ skill: offload-to-sandbox node to the ecosystem graph in vercel.md.

bun run validate passes. The CLI flags were checked against vercel sandbox --help from Vercel CLI 62.1.0.

Adds "Offloading work to a Vercel Sandbox": how to use the vercel sandbox CLI as remote workers for heavy builds, test suites, benchmarks and parallel agent experiments. Covers base snapshots, per-task sandboxes, source pushes that keep dependencies, detached long jobs, fair A/B benchmarking inside one VM, sharding, baseline failure triage, auth, a small wrapper script, and cost hygiene. Complements vercel-sandbox, which covers running untrusted code through the SDK.
Keep the skill to what an agent needs to get offloaded work done: drop the benchmarking section and related signals, keeping one note about comparing timings in the same sandbox, and remove pitfalls that repeat earlier sections.
@cramforce
cramforce marked this pull request as ready for review October 10, 2026 19:56
@vercel
vercel Bot requested a review from molebox October 10, 2026 19:56

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant