skills.rest
AmirulAndalib
AmirulAndalibCOMMUNITY
Explore all skills by this author →
GitHub

crabbox

Runs OpenClaw remote validation on Linux, macOS, Windows, and WSL2 via Crabbox and Blacksmith Testbox.

Updated Jun 19, 2026
Search Tags:#cross-platform#pnpm#e2e testing#ci validation#remote testing#crabbox#blacksmith testbox
One-click install
npx skills add https://github.com/AmirulAndalib/Vilvona-AI --skill crabbox-amirulandalib
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: crabbox
Source: https://github.com/AmirulAndalib/Vilvona-AI/tree/main/.agents/skills/crabbox
Command: npx skills add https://github.com/AmirulAndalib/Vilvona-AI --skill crabbox-amirulandalib

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Validating OpenClaw changes across operating systems and CI-parity environments requires remote machines, synced checkouts, secrets handling, and reliable cleanup, which is error-prone to orchestrate by hand. This Skill standardizes how to run broad pnpm gates, E2E bug proofs, and desktop sessions on remote Crabbox leases or delegated Blacksmith Testbox runners while reporting the actual provider and lease id. ## Core Features & Use Cases - Provider-aware remote runs: Choose between brokered AWS Crabbox (cbx_ ids) and delegated Blacksmith Testbox (tbx_ ids) for broad pnpm check/test gates, with correct flags for each backend. - Cross-platform targets: Execute validation on Linux, native Windows, WSL2, macOS, and static SSH hosts, including desktop/WebVNC sessions for human QA handoff. - E2E bug verification workflow: Reproduce user-facing failures from real entrypoints (install, Docker, gateway, CLI), then prove the fix with one remote run and recorded evidence. - Use Case: A maintainer fixes a packaging bug and needs proof: run the Docker/package E2E lane on a Testbox through Crabbox with --timing-json, capture the tbx_ id and Actions run URL, and report redacted results in the PR. ## Quick Start Ask the agent to run the changed-files test gate remotely through Crabbox with the Blacksmith Testbox provider and report the lease id and timing summary.

Frequently Asked Questions about crabbox

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run OpenClaw pnpm tests on a remote Linux box with Crabbox?▼

Run pnpm crabbox:run with --timing-json and a shell command like pnpm test:changed from the repo root. The wrapper syncs your checkout to a brokered AWS lease and returns a JSON summary with the cbx_ lease id, phases, and exit code.

What is the difference between AWS Crabbox and Blacksmith Testbox?▼

Brokered AWS Crabbox is the direct provider with cbx_ lease ids and syncDelegated=false. Blacksmith Testbox runs through Crabbox as a delegated provider with tbx_ ids, where the Testbox workflow owns checkout, setup, and secrets.

Can Crabbox run tests on Windows or macOS?▼

Yes. Crabbox supports native Windows, WSL2, and macOS targets, including static SSH hosts via --provider ssh --static-host. Broad Linux gates should stay on Linux or Testbox unless the bug is platform-specific.

How do I handle secrets when testing live providers in Crabbox?▼

Use --allow-env or --env-from-profile to forward only the exact needed key into the remote process for that one command. Never print secret values, sync them as repo files, or leave them in remote shell history; delegated Testbox runs require secrets in the workflow instead.

What should I do when a Crabbox lease or sync looks stale?▼

Retry with --full-resync (alias --fresh-sync) before replacing the lease; it resets the remote workdir and re-uploads the checkout. If the run failed or was interrupted, verify cleanup with crabbox list --provider aws or blacksmith testbox list --all.

When should I not use Blacksmith Testbox for validation?▼

Avoid Testbox when you need direct AWS behaviors like --fresh-pr, --full-resync, capture/download flags, or environment forwarding, since delegated providers reject these. Also stay on brokered AWS when Blacksmith is down or quota-limited, and report the outage.

Related Skills

Explore More Software Engineering Skills→
openclawopenclaw
388.4k

openclaw-testing

Route, run, and debug OpenClaw tests, CI lanes, and release validation.

Official
Advanced
vut08905vut08905

openclaw-pr-maintainer

Automate GitHub PR triage by applying labels and safe comments.

Community
Advanced
OmniNode-aiOmniNode-ai
3

runner

Manage GitHub Actions self-hosted runner deployment, updates, and health monitoring.

Official
Advanced
BitRaptorsBitRaptors
16

archie-deep-scan

Generate a semantic architecture blueprint with per-folder context files and health metrics.

Official
Advanced