Release Preflight
Check whether a skill, repo, or export bundle is clean enough to publish.
Core workflow
- 1. Identify the target path and target type (
skill, repo, or bundle). - Check export-safety risks first: private directories, local-only artifacts, and unnecessary publish surface.
- Check readiness basics for the chosen target type.
- Scan text files for obvious local identity leakage.
- Produce a decision:
ready, ready_after_fixes, or not_ready.
Read references as needed
- - Read
references/rules.md for the rule set and severity model. - Read
references/report-format.md for the report structure and decision meanings. - Read
references/target-types.md for type-specific expectations and minimal publish surfaces. - Read
references/export-safety.md for the default private-path and local-artifact patterns. - Read
references/release-minimal.md before packaging or publishing so the first public surface stays minimal.
Use scripts as needed
- - Use
scripts/release_preflight.py <target-path> [--type ...] [--publish-target ...] to run the preflight check.
Operating rules
- - Prefer blocking obvious private artifacts over guessing intent.
- Prefer suggesting a smaller public surface instead of trying to auto-fix files.
- Treat identity leakage conservatively when absolute local paths or local usernames appear in public-facing text files.
- Keep the first version focused on text reports and P0 checks; do not expand into full secret scanning or automated publishing.
发布预检
检查技能、仓库或导出包是否足够干净以进行发布。
核心工作流程
- 1. 识别目标路径和目标类型(技能、仓库或包)。
- 首先检查导出安全风险:私有目录、仅限本地的工件以及不必要的发布范围。
- 检查所选目标类型的基本就绪状态。
- 扫描文本文件是否存在明显的本地身份泄露。
- 生成决策结果:就绪、修复后即可就绪或未就绪。
根据需要查阅参考资料
- - 查阅 references/rules.md 了解规则集和严重性模型。
- 查阅 references/report-format.md 了解报告结构和决策含义。
- 查阅 references/target-types.md 了解特定类型的期望和最小发布范围。
- 查阅 references/export-safety.md 了解默认的私有路径和本地工件模式。
- 在打包或发布前查阅 references/release-minimal.md,确保首次公开范围保持最小。
根据需要运行脚本
- - 使用 scripts/release_preflight.py <目标路径> [--type ...] [--publish-target ...] 运行预检检查。
操作规则
- - 优先阻止明显的私有工件,而非猜测意图。
- 优先建议缩小公开范围,而非尝试自动修复文件。
- 当绝对本地路径或本地用户名出现在面向公众的文本文件中时,对身份泄露采取保守处理。
- 保持首个版本专注于文本报告和P0检查;不要扩展到完整的秘密扫描或自动发布。