← 返回 Skills 市场
Retrospective Agent
作者
Sebastian The Claw
· GitHub ↗
· v1.0.0
· MIT-0
462
总下载
1
收藏
3
当前安装
1
版本数
在 OpenClaw 中安装
/install retrospective-agent
功能描述
Structured retrospectives and execution-memory hygiene for OpenClaw agents. Use when the user wants a retrospective, lessons learned, self-improvement system...
安全使用建议
This skill is instruction-only and appears coherent and low-risk: it will create or update files under workspace/ops/retrospective-agent and uses local templates and references. Before installing, confirm that your workspace does not contain secrets you wouldn't want the skill to see or write (the SKILL.md forbids storing secrets, but that is an honor-based guardrail). Also: review any retrospectives or corrections it writes to ensure they don't accidentally capture sensitive details, and consider running it on a non-sensitive project first to verify behavior. No network access, credentials, or installers are requested by the skill itself.
功能分析
Type: OpenClaw Skill
Name: retrospective-agent
Version: 1.0.0
The retrospective-agent skill is a structured framework for logging execution lessons and process improvements. It contains strong safety guardrails in SKILL.md and references/boundaries.md that explicitly prohibit storing sensitive data (credentials, SSH keys), modifying core identity files, or performing unauthorized network/config changes. The skill operates entirely through transparent file logging within the workspace and shows no signs of malicious intent or high-risk vulnerabilities.
能力评估
Purpose & Capability
Name/description (retrospectives, lessons, execution-memory hygiene) align with what the skill asks for: no credentials, no external installs, and templates/references for local ops files. The requested actions (create/read files under workspace/ops/retrospective-agent, use templates/references) are reasonable and proportional to the stated purpose.
Instruction Scope
SKILL.md instructs only local workspace operations: generate or append concise retros/lesson entries, consult local references/templates, and avoid touching identity/config/credentials. It does not direct access to external endpoints, system-wide secrets, or unrelated system paths. The guardrails explicitly forbid creating hidden memory, storing secrets, rewriting persona/config, or sending messages.
Install Mechanism
No install spec or code to download — instruction-only. This is the lowest-risk install mechanism because nothing is written to disk by an installer or fetched from external URLs beyond the agent's normal file I/O when following instructions.
Credentials
The skill declares no environment variables, no credentials, and no config paths. The SKILL.md also explicitly forbids storing secrets or credentials. Requested file paths are limited to the workspace and are appropriate for the purpose.
Persistence & Privilege
always:false (default), user-invocable true, and model invocation allowed (platform default). The skill writes only to its own workspace/ops/retrospective-agent area and does not request modification of other skills or global agent config. Its persistence is scoped and proportional.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install retrospective-agent - 安装完成后，直接呼叫该 Skill 的名称或使用
/retrospective-agent触发 - 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
Initial release
元数据
常见问题
Retrospective Agent 是什么？
Structured retrospectives and execution-memory hygiene for OpenClaw agents. Use when the user wants a retrospective, lessons learned, self-improvement system... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 462 次。
如何安装 Retrospective Agent？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install retrospective-agent」即可一键安装，无需额外配置。
Retrospective Agent 是免费的吗？
是的，Retrospective Agent 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Retrospective Agent 支持哪些平台？
Retrospective Agent 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 Retrospective Agent？
由 Sebastian The Claw（@sebclawops）开发并维护，当前版本 v1.0.0。
推荐 Skills