← 返回 Skills 市场
Parallel AI search
作者
Tristan Manchester
· GitHub ↗
· v1.0.3
· MIT-0
1830
总下载
1
收藏
4
当前安装
4
版本数
在 OpenClaw 中安装
/install parallel-ai-search
功能描述
Use Parallel's parallel-cli to do live web search, URL extraction (clean markdown), deep research reports, bulk data enrichment (CSV/JSON), FindAll entity di...
安全使用建议
This instruction-only skill largely does what it says: it runs the Parallel CLI to search, extract, enrich, and monitor the web. Before installing or running it: (1) Verify the PARALLEL_API_KEY requirement — confirm you want to give the Parallel service an API key and check its scope/permissions. (2) Prefer installing via pipx or inspecting the install script before running curl | bash from https://parallel.ai/install.sh. (3) Be aware outputs are written to /tmp (avoid exposing sensitive data there). (4) If you plan to use Monitor webhooks, only supply webhook URLs you control/trust to avoid data exfiltration. (5) Note the registry metadata omission (it lists no required env vars) — ask the publisher or registry maintainer to correct the declared requirements if you need stricter auditing. If you are uncomfortable with running an unattended install script or giving an external API key, run the CLI in a sandboxed environment or decline installation.
功能分析
Type: OpenClaw Skill
Name: parallel-ai-search
Version: 1.0.3
The skill provides a master interface for the Parallel AI CLI (`parallel-cli`) to perform web search, data extraction, and research. It is classified as suspicious due to the inclusion of a `curl | bash` installation command (`https://parallel.ai/install.sh`) in `SKILL.md` and `references/troubleshooting.md`, which is a high-risk pattern for remote code execution. While the functionality and broad bash permissions appear aligned with the stated purpose of integrating with the Parallel AI service, the instruction to download and execute a remote script represents a significant security risk without clear evidence of malicious intent.
能力评估
Purpose & Capability
The SKILL.md clearly implements a CLI-based 'Parallel' search/extract/enrich workflow that matches the skill name and description. However, registry-level metadata lists no required environment variables while the SKILL.md (and its openclaw metadata) names PARALLEL_API_KEY as the primaryEnv — this mismatch is unexplained and should be clarified.
Instruction Scope
Runtime instructions are focused on the declared capabilities: calling parallel-cli for search, extract, research, enrich, findall, and monitor. The instructions do not ask the agent to read unrelated system files or other credentials. They do suggest writing outputs to /tmp and optionally configuring webhooks (user-provided URLs) — both expected for these tasks but worth noting as data-handling points.
Install Mechanism
The skill has no formal install spec in the registry, but the SKILL.md instructs the agent to run a remote install script via curl -fsSL https://parallel.ai/install.sh | bash (and offers pipx as a fallback). Executing a remote shell script is higher-risk than a registry-managed package; the domain appears to be the official project, which reduces concern, but users should prefer vetted installation methods (pipx/install from known package sources) or inspect the script before running it.
Credentials
The CLI requires authentication (PARALLEL_API_KEY or interactive login), which is proportionate to its web-search/enrichment function. The problem is the package metadata at the registry level declares no required env vars while the skill itself declares PARALLEL_API_KEY as primaryEnv — this inconsistency could mislead users about what credentials will be requested. No unrelated secrets are requested in the instructions.
Persistence & Privilege
The skill does not request always:true and does not ask to modify other skills or system-wide settings. It uses temporary files under /tmp for outputs, which is normal. Autonomous invocation is permitted (platform default) but is not combined with elevated or unexpected privileges here.
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 parallel-ai-search - 安装完成后，直接呼叫该 Skill 的名称或使用
/parallel-ai-search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
v1.0.3
**Major Update: Now uses parallel-cli instead of Node scripts. All previous script-based files removed; new CLI-focused docs and references added.**
- Replaced Node.js script-based workflow with unified `parallel-cli` command-line interface.
- Added new routing and usage instructions for Search, Extract, Deep Research, Enrich, FindAll, and Monitor modes.
- Removed all Node.js scripts and their references.
- Added command templates and troubleshooting guides specific to CLI workflow.
- Documentation now guides users to install and authenticate with `parallel-cli` before use.
v1.0.2
- Replaced helper file `scripts/_lib.mjs` with `scripts/lib.mjs` for better compatibility with installers.
- Updated documentation to warn that underscore-prefixed script helpers may be excluded by some install tools.
- Incremented skill version to 1.1.1.
v1.0.1
**v1.0.1 Changelog — Reworked skill documentation for clarity and focus**
- Rewrites SKILL.md for improved clarity, user guidance, and conciseness.
- Adds practical usage recipes for domain/date-scoped research and pipeline search-extract.
- Specifies Node.js/environment/API prerequisites up front.
- Moves detailed configuration and API reference to references/ subdocs.
- Clarifies troubleshooting steps and output conventions.
v1.0.0
Initial release of Parallel AI Search skill.
- Enables web search and URL content extraction using Parallel Search/Extract APIs.
- Supports up-to-date research, domain-scoped searches, and extraction of clean excerpts/markdown from web and PDF sources.
- Provides executable Node.js scripts for search, extract, and combined workflows via OpenClaw.
- Requires a PARALLEL_API_KEY for use; detailed configuration instructions included.
- Includes best-practice prompting tips, output handling conventions, and error handling guidance.
元数据
常见问题
Parallel AI search 是什么？
Use Parallel's parallel-cli to do live web search, URL extraction (clean markdown), deep research reports, bulk data enrichment (CSV/JSON), FindAll entity di... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 1830 次。
如何安装 Parallel AI search？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install parallel-ai-search」即可一键安装，无需额外配置。
Parallel AI search 是免费的吗？
是的，Parallel AI search 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Parallel AI search 支持哪些平台？
Parallel AI search 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 Parallel AI search？
由 Tristan Manchester（@tristanmanchester）开发并维护，当前版本 v1.0.3。
推荐 Skills