← 返回 Skills 市场
104
总下载
0
收藏
1
当前安装
7
版本数
在 OpenClaw 中安装
/install outclaw-research
功能描述
Deeply research a specific person or organisation for B2B outreach. Pulls from every outreach-relevant tool in the user's inventory (Leadbay/LeadClaw, Linked...
安全使用建议
Before installing or enabling this skill: 1) Verify the origin and trustworthiness of the repository (the skill references scripts under ~/.openclaw/skills/outclaw/ and will attempt to execute them). Inspect any scripts at those paths and confirm they do only what you expect. 2) Confirm which external accounts/tools (Leadbay/LinkedIn/Gmail/OCR services) you will allow the agent to use; the skill does not declare required credentials but will behave differently depending on which tools are connected. 3) Be aware the skill will read your local KB (kb/me/*.md), other local caches, and write persistent profiles containing PII — audit the intended storage location (~/.openclaw/outclaw/kb/) and access controls before use. 4) If you cannot or do not want to grant filesystem/script execution rights, run the skill in a sandboxed environment or deny access to the referenced scripts/tools. 5) Ask the skill author (or repository) to: a) declare required config paths and credentials, b) include or document the helper scripts it invokes, and c) explicitly state retention/consent policies for stored PII. If you want, provide those files or credentials only after manual review.
能力标签
能力评估
Purpose & Capability
The name/description (deep research + persistent KB writes) align with the instructions (query outreach tools, web, local KB and write profiles). However the skill declares no required config paths or credentials even though the runtime steps explicitly read/write ~/.openclaw/outclaw/kb/, call scripts under ~/.openclaw/skills/outclaw/shared/scripts/, and rely on third‑party tools (Leadbay/LinkedIn/etc.). That divergence (no declared paths/creds vs. explicit file/script usage) is a transparency/ownership mismatch.
Instruction Scope
SKILL.md instructs the agent to execute shell/python scripts in the user's home, read and write persistent KB and memory files, call OCR/vision tools, query many external research channels, and check local apps/notes. These actions are within the stated research purpose, but they also entail broad access to local files and potentially sensitive PII. The agent is told to run specific paths (e.g. ~/.openclaw/skills/outclaw/shared/scripts/outclaw_daily_batch.sh, shared/scripts/domain_classifier.py) which may not be part of the package — executing arbitrary local scripts is high-impact if those scripts are replaced or malicious.
Install Mechanism
No install spec (instruction-only), which is lower risk on its face. But the runtime docs expect local helper scripts and utilities to exist in ~/.openclaw/… even though nothing in the skill bundle installs them. That dependency-on-local-scripts without bundling or declaring them is fragile and a risk vector (the skill will try to execute whatever lives at those paths).
Credentials
The skill declares no required env vars or config paths, yet its behavior depends on access to external services (Leadbay/LeadClaw, LinkedIn, Gmail, OCR tools) and local KB files. In practice those tools will require credentials/token access; the skill does not declare these or explain where/which credentials will be used. It will also persist sensitive contact data into the user's KB. The lack of declared credentials/config paths is a transparency gap and makes it harder to reason about what secrets the skill will use or require.
Persistence & Privilege
The skill is designed to create and modify persistent KB pages and tenant memory (explicitly), which is coherent for a research/outreach assistant. It does not set always:true and doesn't request cross-skill config changes. Nonetheless, persistence implies long-term storage of PII (emails, phone numbers, prior conversations), so users should expect permanent writes and review what is stored.
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 outclaw-research - 安装完成后，直接呼叫该 Skill 的名称或使用
/outclaw-research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
v1.0.6
v2.1.33: hard sanity gate refuses truncated Leadbay pulls — caps agent's paraphrase-and-summarize failure mode
v1.0.5
v2.1.32: ensure_leadbay_bound self-heal for plugins.allow + entries + agent skills + extensions dir
v1.0.4
v2.1.31: pull count=50, kill stale --n 3, heredoc-write discipline
v1.0.3
v2.1.30: paste-the-file-verbatim rule + sanity gate kills 'see above' bypass
v1.0.2
v2.1.29: first-action scaffolder rule + chat-reply-must-validate + angle-fit detector (kills 'furthers their sustainability goals' class)
v1.0.1
v2.1.28: agent mindset + options-and-choose + coherence checklist + evidence card + per-tenant style + daily launchd cron
v1.0.0
- Initial release of outclaw-research skill for deep B2B research on people or organizations.
- Integrates multiple outreach-relevant tools (Leadbay/LeadClaw, LinkedIn, Twitter/X, web search, company pages, podcasts, news) to build persistent profiles.
- Supports both bulk lead discovery and targeted research, with clear batch flows and opt-in deep dives.
- Strict "zero-hallucination" policy: every fact must be traceable to a real source; no unverifiable content allowed.
- Results focus: ≥3 recent, specific signals required per profile; research gaps must be reported transparently.
- Comprehensive guidelines for tool usage, data storage, and result presentation to ensure reliability and reproducibility.
元数据
常见问题
Outclaw Research 是什么？
Deeply research a specific person or organisation for B2B outreach. Pulls from every outreach-relevant tool in the user's inventory (Leadbay/LeadClaw, Linked... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 104 次。
如何安装 Outclaw Research？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install outclaw-research」即可一键安装，无需额外配置。
Outclaw Research 是免费的吗？
是的，Outclaw Research 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Outclaw Research 支持哪些平台？
Outclaw Research 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 Outclaw Research？
由 milstan（@milstan）开发并维护，当前版本 v1.0.6。
推荐 Skills