Stats
Actions
Tags
From superpowers
Searches, verifies, and organizes real academic papers with traceable links (DOI, publisher, CNKI, Google Scholar). Handles literature reviews, core-paper lists, and mixed Chinese/English sources.
How this skill is triggered — by the user, by Claude, or both
Slash command
/superpowers:real-literature-traceThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Use this skill to turn a vague literature request into a traceable paper set with clear selection reasons and real source links.
Use this skill to turn a vague literature request into a traceable paper set with clear selection reasons and real source links.
Use it when the user wants any of the following:
When the user wants foreign literature or a mixed Chinese-international set, treat the workflow as two linked passes:
When ranking international papers, prioritize these sources first:
Do not treat a paper as strong just because it appears in Google Scholar. Use Scholar for discovery, then verify with an official source.
If the user does not specify these details, use sensible defaults and state them briefly before searching:
Do not jump straight into broad search if the topic is vague.
If the user says things like:
帮我找文献搜一些论文做文献综述筛选优秀文献给我真实文献地址then narrow the request first.
Ask only the highest-value missing details:
If the user still stays vague after one round, choose one reasonable interpretation, say it explicitly, and proceed.
Follow this order.
Before searching, restate the agreed scope in 2-5 lines:
Prepare a small bilingual keyword set:
Split broad topics into subthemes so the results are diverse rather than repetitive.
If the user wants foreign literature, include venue-aware retrieval terms such as:
Examples:
spacecraft pursuit evasion game IEEE TACorbital pursuit evasion game AIAAdifferential game spacecraft AutomaticaPrefer sources that can be verified later:
For international screening, treat venue quality as part of the retrieval plan:
Do not promote a paper to the final set unless you can explain where it came from and how to check it.
Keep papers that best satisfy most of these conditions:
Reject papers that are:
For each selected paper, return the best available address in this order:
If a link cannot be confirmed, say 待核验 instead of guessing.
Return the final papers with:
strong, selected, or backupUse a practical ranking approach:
A for core papers that are highly relevant and easy to verifyB for good supporting papersC for backup papers that may still be usefulS for standout international papers from top journals or top conferences that are both highly relevant and strongly influentialWhen deciding between similar papers, prefer the one with:
Do not overvalue citation counts alone. A newer but more exact paper can be better than a famous but generic one.
When the user asks for a paper list, use a table by default.
Recommended columns:
If the user wants a literature review or related-work section, summarize the final set into:
npx claudepluginhub lunartech-x/superpowers --plugin superpowersSearches academic literature across multiple sources (Google Scholar, Consensus, Paperpile) with deduplication, DOI resolution, and journal filtering.
Finds research papers answering a query using Firecrawl: semantic search, related-paper expansion, and in-body verification. Supports single-paper lookups and full multi-paper sets.
Runs a systematic literature review workflow: scope definition, multi-source search (arXiv, Semantic Scholar, Google Scholar), screening, extraction, synthesis, and gap analysis.