Transparent, rigorous research with full methodology — not a black-box API wrapper. Conducts exhaustive investigation through mandated 2-cycle research per theme, APA 7th citations, evidence hierarchy, and 3 user checkpoints. Self-contained using native OpenClaw tools (web_search, web_fetch, sessions_spawn). Use for literature reviews, competitive intelligence, or any research requiring academic rigor and reproducibility.
Security Analysis
medium confidenceThe skill's instructions, required tools, and lack of external installs/credentials are consistent with its stated purpose as a self-contained research methodology, with only a few minor documentation/instruction mismatches to note.
The skill claims to be self-contained and 'works offline / no cloud services' in README, but its runtime instructions require web_search and web_fetch (networked web access) — this is not a credentials or install mismatch but is a documentation contradiction. Otherwise the requested tools (web_search, web_fetch, sessions_spawn, memory_search/get) are appropriate for exhaustive, reproducible research.
SKILL.md mandates explicit multi-cycle research, shows tool usage (web_search, web_fetch, sessions_spawn) and requires showing analysis after every tool call; all of that stays within the stated research/literature-review scope. It does instruct checking memory (memory_search/memory_get) for prior context — reasonable for continuity but it means the skill will access agent memory if available.
Instruction-only skill with no install spec, no downloads, and no third-party packages. This is low-risk and proportionate for the described purpose.
The skill requests no environment variables, no credentials, and no config paths. The only sensitive access implied is to the agent's memory (via memory_search/memory_get) and to the web (via web_search/web_fetch), both of which are coherent with research tasks but should be considered when working with sensitive topics.
always is false and model invocation is allowed (platform default). The skill requires multi-step autonomous execution in Phase 3 once approved, which is expected for an automated research workflow. It does not request elevated persistence or modify other skills.
Guidance
This skill appears to do what it says: a strict, multi-cycle web-based research methodology with checkpoints. Before installing or running it, consider: (1) Despite a README claim about 'offline' operation, the skill uses web_search/web_fetch (it will browse the web). If you require truly offline research, do not use it. (2) The skill may read agent memory (memory_search/memory_get) to cross-reference prior context — if your stored memory contains sensitive data, be cautious or disable memory for the session. (3) Phase 3 auto-executes full research cycles after you approve the plan — expect extended web activity and long outputs; review and approve the Phase 2 plan carefully to limit scope, sources, and freshness parameters. (4) Because it is instruction-only and has no installs or credentials, it does not write new code to disk or request secrets, which reduces surface risk. If any of the above concerns matter (sensitive topics, regulatory data, need for offline-only), either restrict the skill's access or decline to use it.
Latest Release
v1.0.0
Initial release: Transparent, rigorous research with full methodology. APA citations, evidence hierarchy, 3 user checkpoints, no external dependencies.
More by @kesslerio
Published by @kesslerio on ClawHub
