← 返回 Skills 市场
Academic Deep Research
作者
MarjorieBroad
· GitHub ↗
· v1.0.0
· MIT-0
57
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install mar-academic-deep-research
功能描述
Transparent, rigorous research with full methodology — not a black-box API wrapper. Conducts exhaustive investigation through mandated 2-cycle research per t...
安全使用建议
This skill appears to do what it says (multi-cycle web research) and is low-risk to install because it has no install steps or credential requests. Before enabling it, consider: (1) The author claims it "works offline" but the instructions call web_search/web_fetch — expect network activity and many outgoing requests. (2) The skill will read platform memory (memory_search/memory_get) if available — remove or audit sensitive memories you don't want included. (3) After you approve the research plan the skill will run fully and autonomously for that job (many requests and long outputs); if you need finer control, decline autonomous execution or require repeated approvals. (4) The skill mandates APA citations and 1–2 citations per paragraph — ensure the platform preserves verifiable source links because the skill can otherwise produce plausible but fabricated citations. If you rely on this for regulated or sensitive topics (healthcare, legal, proprietary IP), test with non-sensitive queries first and confirm how web_fetch results and memory content are logged, retained, and shared by the platform.
功能分析
Type: OpenClaw Skill
Name: mar-academic-deep-research
Version: 1.0.0
The 'academic-deep-research' skill bundle is a highly structured prompt engineering template designed to guide an AI agent through a rigorous, multi-phase research process. It utilizes standard OpenClaw platform tools (web_search, web_fetch, sessions_spawn) for their intended purposes of data gathering and parallel processing. The instructions in SKILL.md emphasize transparency, academic rigor (APA citations), and user control through three mandatory checkpoints, which actually enhances safety rather than circumventing it. No evidence of data exfiltration, malicious execution, or harmful prompt injection was found; the logic is entirely consistent with the stated goal of providing deep, reproducible research.
能力评估
Purpose & Capability
The skill claims to be a self-contained, reproducible research assistant and its runtime instructions use only platform tools (web_search, web_fetch, sessions_spawn, memory_get). That aligns with the stated purpose. However, the README asserts 'Works offline — No third-party API keys required' which contradicts explicit use of web_search/web_fetch (network I/O). This is an internal inconsistency a user should be aware of.
Instruction Scope
SKILL.md stays focused on research: multi-cycle web searches, fetches, synthesis, and mandatory user approvals for planning. It does instruct use of memory_search/memory_get and to 'Check MEMORY.md for related context' — which means the agent will pull prior user memories or stored context. That could surface private or sensitive user data if present. The skill also mandates showing analysis after every tool call and performing two full cycles per theme, which will result in many external requests and lots of aggregated output; this is expected for the stated purpose but increases the potential for unintended disclosure or heavy resource use.
Install Mechanism
This is an instruction-only skill with no install spec, no added binaries, and no code files to execute on disk. From an install-mechanism perspective there is nothing being downloaded or installed, which minimizes supply-chain risk.
Credentials
The skill declares no required environment variables, no credentials, and no config paths. That is proportionate to its described web-research purpose. Note: it does request access to platform memory APIs (memory_search/memory_get) which is not an environment variable but is a data source; that access is reasonable for contextual research but can expose stored memories.
Persistence & Privilege
The skill is not marked always: true and uses the default autonomous-invocation capability. Autonomous invocation is platform-default and not flagged here by itself. Be aware Phase 3 (execution) is 'NO STOPS — EXECUTE FULLY' after the Phase 2 user approval, so once the plan is approved the skill will run many steps without further confirmation. That increases the blast radius if the research topic or memory context is sensitive.
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 mar-academic-deep-research - 安装完成后，直接呼叫该 Skill 的名称或使用
/mar-academic-deep-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
v1.0.0
- Initial release of the Academic Deep Research skill for transparent, rigorous, and reproducible investigations.
- Implements mandated two-cycle research per theme, with explicit evidence trails and contradiction documentation.
- Structured workflow with three user checkpoints: clarifying questions, research plan approval, and final synthesis.
- Utilizes only SkillBoss platform-native tools (web_search, web_fetch, sessions_spawn) for all research tasks.
- Emphasizes APA 7th citation style, evidence hierarchy, and analysis methodology suitable for literature reviews and academic-quality reports.
- Integrated protocol for handling insufficient data, unresolved contradictions, and source quality concerns.
元数据
常见问题
Academic Deep Research 是什么？
Transparent, rigorous research with full methodology — not a black-box API wrapper. Conducts exhaustive investigation through mandated 2-cycle research per t... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 57 次。
如何安装 Academic Deep Research？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install mar-academic-deep-research」即可一键安装，无需额外配置。
Academic Deep Research 是免费的吗？
是的，Academic Deep Research 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Academic Deep Research 支持哪些平台？
Academic Deep 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 Academic Deep Research？
由 MarjorieBroad（@marjoriebroad）开发并维护，当前版本 v1.0.0。
推荐 Skills