← 返回 Skills 市场
134
总下载
0
收藏
0
当前安装
8
版本数
在 OpenClaw 中安装
/install workspace-local-retrieval
功能描述
Build boundary-first local retrieval for OpenClaw with explicit corpora, deny-by-default agent access, separate memory layers, and a validated minimal demo p...
安全使用建议
This skill appears coherent and conservative, but review and confirm before running: (1) The bootstrapper will write template JSON files to the destination you provide — do not run with --force unless you want overwrites. (2) The skill does not auto-index or call the network, but the default backend template points to a local Ollama endpoint (127.0.0.1) — verify or change that before enabling embeddings. (3) If you plan to install the CLI via npm, prefer installing from a trusted package source; installing arbitrary global packages carries supply-chain risk. (4) Run the prereq checker first (workspace-local-retrieval check) and inspect generated config files before performing indexing or connecting any embedding backend. (5) If you need greater assurance, ask the publisher for an official release URL or audit the package that would appear on npm before installing.
功能分析
Type: OpenClaw Skill
Name: workspace-local-retrieval
Version: 1.0.0
The skill bundle provides a robust, privacy-focused framework for bootstrapping local Retrieval-Augmented Generation (RAG) architectures within an OpenClaw workspace. It includes a Node.js CLI (bin/workspace-local-retrieval.js) and Python scripts for environment validation (check_retrieval_prereqs.py) and configuration template generation (bootstrap_workspace_retrieval.py). The architecture emphasizes 'deny-by-default' access policies and explicit data boundaries to prevent accidental exposure of sensitive information. The code is well-structured, lacks network calls or data exfiltration logic, and the instructions in SKILL.md promote safe agent behavior by requiring explicit user authorization and a task plan before making any environment changes.
能力评估
Purpose & Capability
Name/description match the included artifacts: a CLI wrapper, prereq checker, and bootstrapper that generate sanitized config templates for a local retrieval stack. Required capabilities (Python, Node, SQLite, optional local Ollama) are consistent with a local-first retrieval purpose.
Instruction Scope
SKILL.md and scripts constrain actions to: running prerequisite checks, writing template config files, and recommending next steps. The bootstrapper explicitly avoids network calls and does not index or scan user data. The CLI invokes only the included Python scripts and performs local file writes; no instructions ask the agent to read unrelated secrets or transmit data.
Install Mechanism
There is no registry install spec in the manifest (the skill is instruction/codeshipped), but package.json and bin/ provide an npm-friendly CLI. This is not harmful, but users should note that the npm-install claim in SKILL.md presumes publishing to npm; installing arbitrary packages from unknown registries has typical supply-chain risks.
Credentials
The skill declares no required env vars or credentials. The code respects optional local services (ollama) and uses PYTHON env var as an override for which Python executable to run — reasonable and proportionate for cross-platform scripts. No unrelated credentials are requested or used.
Persistence & Privilege
The skill does not request always:true or elevated platform privileges. It writes only template config files under the user-specified destination (refuses to overwrite without --force) and does not modify other skills or system-wide agent settings. Autonomous invocation is allowed by default (normal) but the skill's runtime behavior is local and conservative.
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 workspace-local-retrieval - 安装完成后，直接呼叫该 Skill 的名称或使用
/workspace-local-retrieval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
Release 1.0.0: npm-friendly CLI bootstrapping with prerequisite checks and updated documentation
v0.3.4
Add OS-specific prerequisite installers, a one-command demo setup path, a real-workspace quickstart, and clearer GitHub-facing positioning for validated minimal demo vs non-turnkey workspace retrieval.
v0.3.3
Rename display name to Agent-Scoped Local Retrieval for clearer external positioning; keep slug unchanged for link stability.
v0.3.2
Shorten ClawHub-facing copy and summary for faster external reading while keeping boundary-first positioning and validation claims precise.
v0.3.1
Polish ClawHub-facing positioning and summary for external readers; tighten frontmatter description and listing copy while preserving validation honesty.
v0.3.0
Add validation contract and anti-overclaim guidance; align publish wording with maturity rules; add minimal closed-loop demo corpus, SQLite FTS5 index/search scripts, and passing smoke tests for a fully validated minimal demo path.
v0.2.0
Add explicit prerequisite gate, OS-aware install policy, dependency guidance, backend config template, and runnable environment checks. Skill now blocks execution when required dependencies are missing and can first drive installation planning.
v0.1.0
Initial public release: local-first retrieval architecture with explicit corpus boundaries, deny-by-default agent access, sanitized starter templates, design rationale, and maintenance-aware refresh guidance.
元数据
常见问题
Workspace Local Retrieval 是什么？
Build boundary-first local retrieval for OpenClaw with explicit corpora, deny-by-default agent access, separate memory layers, and a validated minimal demo p... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 134 次。
如何安装 Workspace Local Retrieval？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install workspace-local-retrieval」即可一键安装，无需额外配置。
Workspace Local Retrieval 是免费的吗？
是的，Workspace Local Retrieval 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Workspace Local Retrieval 支持哪些平台？
Workspace Local Retrieval 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 Workspace Local Retrieval？
由 YS-c-23（@ys-c-23）开发并维护，当前版本 v1.0.0。
推荐 Skills