← 返回 Skills 市场
128
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install openclaw-read-flow
功能描述
基于 FreshRSS 的漏斗式阅读流总技能:从未读抓取、昨天窗口切片、Digest 构建,到固定 6 栏 Daily Review 和后续知识沉淀。用于需要把聚合信息流加工成可读日报与可复用知识资产的场景。
安全使用建议
This skill's instructions expect FreshRSS credentials and local config access but the metadata does not declare any required credentials — that mismatch is the main concern. Before installing or using it: (1) Confirm how credentials will be provided and avoid giving full account credentials; create a limited API user if possible. (2) Be cautious about letting the agent read local config files (they may contain sensitive secrets). (3) Note that the skill ships no code; the agent will be following prose and referenced external scripts that are not provided — verify those scripts exist and review them. (4) Test the flow in a sandbox or with a disposable FreshRSS account and ensure the agent does not automatically write back (mark read or push to long-term stores) unless you explicitly enable that. (5) If you need stronger assurance, ask the publisher to update the manifest to declare required env vars and to provide audited scripts or code so behavior is explicit.
功能分析
Type: OpenClaw Skill
Name: openclaw-read-flow
Version: 1.0.0
The skill bundle defines a structured workflow for processing FreshRSS feeds into digests and daily reviews. The instructions in SKILL.md and workflow.md are well-defined, focusing on content deduplication, Chinese summarization, and categorization into specific columns. While the skill handles FreshRSS credentials and references external scripts (e.g., in ../digest-builder/scripts/), these are necessary for the stated functionality. The instructions explicitly include safety constraints, such as forbidding the agent from automatically marking items as read or writing to long-term knowledge bases without human confirmation, which demonstrates a focus on user control rather than malicious autonomy.
能力评估
Purpose & Capability
The skill's stated purpose (pull unread items from FreshRSS, build digests, produce Daily Review) is coherent, but the SKILL.md explicitly requires access to FreshRSS configuration (base_url, username, api_password) and recommends scripts to run. The registry metadata declares no required credentials or binaries. Requiring FreshRSS credentials is reasonable for the stated purpose, but those requirements are not reflected in the skill manifest, which is an inconsistency.
Instruction Scope
The instructions direct the agent to read FreshRSS config files and environment variables, call the FreshRSS Google Reader API, fetch article bodies, deduplicate, and produce structured outputs. Reading configuration files (to obtain api_password) is explicitly suggested; the skill also references local script paths and directories. There are no instructions to exfiltrate data to third-party endpoints, but the directive to read local config (which may contain sensitive credentials) is broader than the manifest indicates and could expose secrets if the agent is given broad file access.
Install Mechanism
This is an instruction-only skill with no install spec and no included code files. That minimizes install-time risk (nothing is downloaded or written by an installer). The recommended scripts are referenced but not packaged, so use requires external tooling that the agent or user must provide.
Credentials
The SKILL.md expects FreshRSS credentials (base_url, username, api_password) and prioritizes reading config files and env vars, but the skill metadata lists no required environment variables or primary credential. Asking for access to account credentials is proportionate to the purpose, but omitting them from the declared requirements is a red flag: it makes it unclear how the agent will obtain secrets and whether secrets might be read from unexpected locations.
Persistence & Privilege
The skill is not always-enabled and allows user invocation; disable-model-invocation is false (normal). It does not request permanent presence, nor does it declare actions that modify other skills or system-wide settings. The SKILL.md states default behavior not to mark items as read or write to long-term knowledge without confirmation, which limits persistence risk — but that is only descriptive (no code enforces it).
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 openclaw-read-flow - 安装完成后,直接呼叫该 Skill 的名称或使用
/openclaw-read-flow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 of the unified OpenClaw read flow skill.
元数据
常见问题
openclaw-read-flow 是什么?
基于 FreshRSS 的漏斗式阅读流总技能:从未读抓取、昨天窗口切片、Digest 构建,到固定 6 栏 Daily Review 和后续知识沉淀。用于需要把聚合信息流加工成可读日报与可复用知识资产的场景。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 128 次。
如何安装 openclaw-read-flow?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install openclaw-read-flow」即可一键安装,无需额外配置。
openclaw-read-flow 是免费的吗?
是的,openclaw-read-flow 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
openclaw-read-flow 支持哪些平台?
openclaw-read-flow 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 openclaw-read-flow?
由 阿晨聊技术(@gcdd1993)开发并维护,当前版本 v1.0.0。
推荐 Skills