← 返回 Skills 市场
gavinlinasd

Self Preserve

作者 Gavin Lin · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ✓ 安全检测通过
270
总下载
1
收藏
0
当前安装
8
版本数
在 OpenClaw 中安装
/install self-preserve
功能描述
Backup readiness and disaster recovery for your OpenClaw agent. Checks whether your memory, identity, config, skills, and workspace files are covered by a re...
安全使用建议
This skill is internally consistent and low-risk for its stated purpose, but review these before installing: (1) it will run ls -la on files under ~/.openclaw/ and ~/openclaw-backups/ — filenames and metadata (not contents) will be observed; avoid encoding secrets into filenames. (2) it may read an optional ~/.openclaw/offsite.json marker file — do not store credentials or sensitive endpoints there. (3) if you allow cron scheduling, persistent schedules will be written to the agent-managed .claude/scheduled_tasks.json; only confirm if you trust the schedule being created. The skill does not perform offsite uploads — you must implement remote backup yourself. If you want to be extra cautious, run the skill in a restricted environment or inspect the SKILL.md content yourself before use.
能力评估
Purpose & Capability
Name/description (backup/readiness/disaster recovery) match the actual instructions: the skill only lists standard OpenClaw state files, checks for local backup archives, and can schedule cron jobs using agent-native CronCreate/CronList/CronDelete. It requests no env vars, binaries, or installs, which is appropriate for this purpose.
Instruction Scope
Instructions restrict operations to ls -la on allowlisted paths (~/.openclaw/ and ~/openclaw-backups/) and explicit denylisting of credentials directories. This is coherent, but two caution points remain: (1) filenames/sizes/timestamps can still leak information if secrets are embedded in filenames, and the skill does read a user-managed marker (~/.openclaw/offsite.json) which could contain sensitive metadata if the user put it there. The safety relies on the agent following the 'never output secrets' rule and the user not storing secrets in file names or the marker file.
Install Mechanism
No install spec and no code files — instruction-only skill — which minimizes install-time risk. Nothing is downloaded or written during installation.
Credentials
The skill declares no required environment variables or credentials and explicitly denylists credential locations (~/.openclaw/credentials/, ~/.ssh/, ~/.aws/, etc.). The declared env/credential footprint is minimal and proportionate to the stated functionality.
Persistence & Privilege
Skill is not forced-always. It may create persistent cron schedules only after explicit user confirmation; persistent schedules are written to the agent's .claude/scheduled_tasks.json (documented). Autonomous invocation is allowed by platform default but is not combined with broad credential access or always:true in this skill.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install self-preserve
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /self-preserve 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
Self-preserve 1.0.0 - Initial release providing backup readiness and disaster recovery assessment for OpenClaw agents. - Checks existence and backup status of agent config, memory, identity, workspace, and skills files. - Detects local backups, surfaces offsite backup marker file if present, and inspects recent backup history. - Advises on best practices for offsite disaster recovery and version control (advisory only, does not implement). - Supports user-gated creation and management of automated backup cron jobs via Claude-native tools. - Strictly read-only for critical files; never reads contents or accesses credential directories.
v0.4.0
**v0.4.0 — Adds offsite backup awareness and guidance** - Now checks whether an offsite copy of your backup exists (declared via an optional marker file), highlighting the gap between local backups and true disaster recovery. - Describes why local-only backups can’t protect against disk loss, device theft, or wipe, and offers advisory steps for offsite/remote backup (e.g., using git remotes, rsync/scp, S3-compatible storage), without automating or vendor-locking any solution. - Report format and next-step recommendations expanded to surface offsite backup status and last confirmed date. - Security and allowlist language updated: now permits read-only access to the offsite marker file, while keeping file writes and network access barred. - No code/scripts or automation; new features remain advisory and user-controlled.
v0.3.2
Self-Preserve 0.3.2 Changelog - Added author and repository fields to SKILL.md for improved attribution and source tracking. - Updated homepage URL to pineappleai.com. - No functional changes to logic or usage; safety rules and backup assessment instructions remain unchanged. - Minor documentation tweaks for consistency.
v0.3.1
Self-preserve 0.3.1 changelog: - Adds advisory to recommend version control and session-end hooks for identity files; the skill itself does not implement these. - Clarifies that advisory recommendations are informational only and no hooks or version control will be configured by the skill. - Now suppresses the display of non-backup cron jobs to the user when checking for automated backups. - Documentation updated for clarity and to reflect recommendations for improved disaster recovery practices.
v0.3.0
Version 0.3.0 of the self-preserve skill introduces automated backup scheduling and updated disaster recovery assessment: - Adds support for creating, updating, and deleting automated backup cron jobs (with user confirmation). - Significantly expands the SKILL.md to clarify security boundaries and integration with Claude-native cron tools (CronCreate, CronList, CronDelete). - Tightens path allowlist and credential exclusion rules; improves wording for security vetting and path safety. - Changes reporting language: the main report now shows "Protected?" status using ⚠ and ✅ for clearer risk signaling. - Updates instructions and recommendations to explicitly guide users toward automated, regular backups.
v0.2.0
Functional backup assessment skill. Agent instructions with explicit safety rules, constrained paths, metadata-only file checks. No credential access, no file content reading, no network calls.
v0.1.1
Remove credential and API key references from SKILL.md to address security scan flag. Skill explicitly declares it does not access any secrets, tokens, or credentials.
v0.1.0
Initial release of the self-preserve skill. - Provides self-assessment for backup readiness, disaster recovery, and state protection of your OpenClaw agent. - Inventories workspace, memory, credentials, skills, and configuration to identify unprotected or at-risk state. - Guides users toward full agent recoverability and highlights what needs backup. - No code execution or network calls; fully instruction-based for secure assessment. - v0.1.0 focuses on workspace inventory and backup status evaluation.
元数据
Slug self-preserve
版本 1.0.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 8
常见问题

Self Preserve 是什么?

Backup readiness and disaster recovery for your OpenClaw agent. Checks whether your memory, identity, config, skills, and workspace files are covered by a re... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 270 次。

如何安装 Self Preserve?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install self-preserve」即可一键安装,无需额外配置。

Self Preserve 是免费的吗?

是的,Self Preserve 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

Self Preserve 支持哪些平台?

Self Preserve 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 Self Preserve?

由 Gavin Lin(@gavinlinasd)开发并维护,当前版本 v1.0.0。

💬 留言讨论