360 Secure SkillHub
360 安全技能中心

Install Calibrate

该工具借助calibrate CLI,为JS/TS应用提供Calibrate浏览器SDK及独立sidecar的安装、配置、验证与排查服务。支持React/Vite等项目,含预检查、多模式工作流、故障恢复机制。典型场景包括添加Calibrate、埋点引导流程、创建清单路由图等,需按流程执行并经用户授权,最终给出对应等级的验证反馈结果。

ojusave
ojusave
热度 15.01
最近更新 2026-09-17 02:25:26
分类:开发依赖:代码执行 / 读写文件应用场景:API 集成 / 调试排错 / 自动化脚本 / DevOps

安装技能

复制下面的话术发给你的 Agent,它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill: npx secure-skills-cli add install-calibrate
Metadata
name
install-calibrate
description
Install, configure, verify, or troubleshoot the Calibrate browser SDK and standalone sidecar in JavaScript or TypeScript applications. Use when a user asks an AI coding agent to add Calibrate, instrument an onboarding flow, create a manifest and route map, run the collector, validate privacy boundaries, or repair an existing Calibrate installation.

Install Calibrate

Use the calibrate CLI for deterministic detection, planning, application, and verification. Inspect the target repository and its local instructions before running it.

Preflight

Before planning, confirm:

  1. Node.js is version 20 or newer and usecalibrate is version 0.1.4 or newer.
  2. The target is a supported React/Vite or generic ESM browser application with one detectable entry point.
  3. The target repository's package manager, current worktree changes, and local instructions are understood.
  4. The collector URL already exists, or the user has chosen the local standalone sidecar path.
  5. No custom Calibrate integration exists that the installer cannot safely reconcile.

Workflow

When the user supplies an existing collector URL and wants the shortest guided flow, run npx usecalibrate install --url <collector-url> --json first without --yes. This preview intentionally exits with code 3 and makes no project changes. Review its collector checks, exact manifest, route mapping, plan.changes[].content, dependency command, required environment names, and dashboard URL. Compare proposed contents with the current files and show the user the exact diffs. After explicit approval, rerun with --yes --json. Provide CALIBRATE_WRITE_KEY through the environment only when collector runtime verification is required.

When the collector URL is unavailable, the mapping needs manual investigation, or a durable plan artifact is useful, use the lower-level workflow:

  1. Run npx usecalibrate detect --dir . --json without changing files.
  2. Review the detected framework, entry point, existing installation, and proposed fixed routes.
  3. If routes are missing or ambiguous, ask the user which ordered routes form the onboarding flow and which route means shipped.
  4. Run npx usecalibrate plan --dir . --out calibrate.plan.json, adding explicit --route values when needed.
  5. Show the user the proposed manifest, collector mode, exact generated contents or diffs, dependency command, required environment names, and unresolved decisions.
  6. Apply only after approval with npx usecalibrate apply --plan <plan-file> --yes.
  7. Run the host project's build or typecheck.
  8. Run npx usecalibrate verify --dir . --json for static checks.
  9. When a local sidecar and write key are available, set CALIBRATE_WRITE_KEY in the environment and run collector runtime verification with --endpoint.
  10. Start the real host application, visit the mapped routes, and confirm an expected dashboard count changes before reporting the application flow as validated.

Read references/cli.md for command details, exit codes, supported projects, and recovery behavior.

Guardrails

  • Use fixed machine identifiers only.
  • Never derive identifiers from form values, DOM text, clipboard contents, URLs, query strings, hashes, or other user input.
  • Keep ADMIN_TOKEN server-only.
  • Treat WRITE_KEY and DASHBOARD_TOKEN as scoped browser-visible workshop credentials, not user authentication.
  • Keep browser and collector manifests identical by using calibrate.install.json as the installer record.
  • Do not overwrite a custom Calibrate integration or ambiguous application entry point.
  • Do not deploy, publish, create accounts, or modify production infrastructure without explicit authorization.
  • Treat --url as an existing collector. The installer does not deploy or host infrastructure.
  • Do not pass secrets as command-line arguments. Command arguments can remain in shell history or process listings.
  • Do not revert changed files automatically after a partial failure. Preserve the failure, show the changed files and failing check, and ask before any rollback.

Failure and recovery

  • Exit code 3: report the unresolved decision or unsupported target. Confirm that no project files changed.
  • Exit code 4 before writes: report the failed check and leave the project unchanged.
  • Exit code 4 after writes: report every changed file, preserve user changes, and show the failed build or verification check. Do not claim completion.
  • Stale plan: discard the plan, rerun detection and planning, and request approval for the new plan.
  • Dependency installation failure: do not retry with a different package manager. Report the command and exit code, then follow the repository's package-manager convention.

Completion

A package install or passing build is not enough. Report exactly one applicable level:

  • Static integration verified: generated files, dependency installation, build or typecheck, and static verification passed.
  • Collector runtime verified: a synthetic completed journey reached the collector and the privacy rejection check passed.
  • Application flow validated: the real host application loaded the SDK, mapped routes were visited, and an expected dashboard count changed.

Never use collector runtime verification as proof that the host application's browser path works.