Transparent, rigorous research with full methodology — not a black-box API wrapper. Conducts exhaustive investigation through mandated 2-cycle research per t...

This skill appears to do what it says: disciplined, multi-cycle web research using OpenClaw native tools and explicit user checkpoints. Before installing or using it, consider the following: (1) Clarify the README claim that it 'works offline' — in practice it uses web_search/web_fetch and will perform many external web requests; it does not require external API keys, but it is not truly offline. (2) If you plan to research sensitive or regulated material (PHI, proprietary code, non-public documents), verify your organization’s policies: web_fetching external sites can transmit query strings and retrieved content back into the agent and possibly logs. (3) sessions_spawn spawns parallel sub-agents that will also perform searches/fetches; confirm what permissions those sub-agents inherit and how their network activity is logged. (4) Expect high volume of web requests (many searches and fetches per theme); test first on non-sensitive topics to observe behavior and resource usage. (5) If you need absolute assurance about data jurisdiction or on-prem processing, validate vendor/platform documentation about where web_search/web_fetch endpoints operate and whether any content is retained. If these operational details are acceptable, the skill is coherent and proportionate to its stated purpose.

Type: OpenClaw Skill Name: academic-deep-research-pro Version: 1.0.0 The academic-deep-research-pro skill bundle is a well-structured framework designed to guide an AI agent through a rigorous, multi-phase research process. It mandates a two-cycle investigation per theme, enforces evidence hierarchy standards, and requires explicit user checkpoints for approval in SKILL.md and reference/protocol.md. The use of native OpenClaw tools like web_search and sessions_spawn is entirely consistent with the stated goal of transparent, reproducible research, and there is no evidence of data exfiltration, malicious execution, or harmful prompt injection.