Release Skills

WHAT: Universal release workflow with auto-detection, multi-language changelogs, semantic versioning, and git tagging. WHEN: User wants to create a release, bump version, update changelog, push a new version, or prepare for deployment. KEYWORDS: "release", "发布", "new version", "新版本", "bump version", "update version", "更新版本", "push", "推送", "create release", "prepare release", "tag version"

This skill contains plausible release automation code but the pieces don't fully match its description. Before installing or running it: 1) Review the three scripts — they operate on .agents/skills/* and modify SKILL.md frontmatter (not necessarily package.json/pyproject/Cargo.toml as the marketing suggests). 2) Ensure git is installed and understand that commits/tags/pushes will be performed; test with --dry-run and in a throwaway clone so you don't accidentally modify or push many SKILL.md files. 3) Backup your repository or use a branch. 4) If you expect a generic project releaser (package.json, pyproject, Cargo.toml), ask the publisher or inspect the code to confirm those paths are actually written/updated. 5) Confirm prompts cannot be bypassed in your usage (so pushes require explicit consent). If you don't want changes to .agents/skills, do not run --all or run the tool pointed at a specific path and examine diff before committing.

Type: OpenClaw Skill Name: release-skills Version: 0.1.0 The OpenClaw AgentSkills skill bundle provides a universal release workflow. The Python scripts (`prepare_release.py`, `release_notes.py`, `roadmap_changelog.py`) interact with Git and modify local markdown files (SKILL.md, CHANGELOG.md) as part of their stated purpose. All external command execution via `subprocess.check_output` uses argument lists, mitigating shell injection risks. The `SKILL.md` explicitly outlines a user confirmation step before pushing changes and includes 'NEVER' rules that reinforce safe behavior, such as never force-pushing or skipping user consent. There is no evidence of data exfiltration, unauthorized network activity, persistence mechanisms, or obfuscation, and all actions align with the described functionality.