Compare local test results with E2B sandbox executions and emit deterministic verdicts.
npx skills add https://github.com/paulingham/.claude --skill sandbox-verify
Please help me install this Agent Skill. Skill: sandbox-verify Source: https://github.com/paulingham/.claude/tree/main/skills/sandbox-verify Command: npx skills add https://github.com/paulingham/.claude --skill sandbox-verify
Gate builds by cross-checking the local test results against a remote sandbox, catching divergences early to prevent shipped defects.
After your local tests pass, trigger sandbox-verify to compare per-test results with the sandbox and emit a final verdict.
High-intent search queries and answers about installing and using this skill.
To verify local test results match sandbox execution, trigger a parity check that provisions a remote E2B sandbox, runs the project tests, and compares per-test outcomes to catch cross-environment divergences before deployment.
Sandbox parity checking for build pipelines is the process of cross-checking local test pass sets against remote sandbox executions to ensure cross-environment consistency and gate deployments against reproducible test outcomes.
Yes, you need an E2B sandbox to run cross-environment parity checks, as the verification process provisions it to execute tests remotely and compare results against your local run for deterministic verdicts.
The deterministic verdicts emitted when comparing local and sandbox test outcomes are SANDBOX_VERIFIED when passes match, SANDBOX_FAILED when they diverge, and SANDBOX_SKIPPED when the sandbox is unavailable.
If the sandbox is unavailable during verification, the build gate process emits a SANDBOX_SKIPPED verdict, allowing the pipeline to handle the skipped state without failing the cross-environment consistency check.
Your CI pipeline needs cross-environment test consistency verification to catch divergences between local and isolated sandbox runs early, preventing shipped defects and ensuring reproducible test outcomes for compliance and reliability.
duany049Execute pilot experiments from YAML specs and collect raw JSON results.
iamtouchskyerCoordinate a multi-agent digraph review pipeline with gate verdicts.
samuerioExecute isolated code and file operations in E2B sandboxes.
nshaikhsCoordinate remote sandboxed agents with per-session state isolation and GitHub authentication.
