agentsclimarketplace

Literature review

Skill Amey-Thakur/AI-SKILLS/skills/research/literature-review

Plug-and-play skills and prompts for every AI coding agent

Install
npx -y skills add Amey-Thakur/AI-SKILLS --skill literature-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 19 days oldThe repository was created 19 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Survey prior work systematically through search strategy, citation chasing, quality assessment, and synthesis. Use when entering a new domain or grounding a decision in what is already known.

SKILL.md

3.5 KB, as published. Nobody here has run it

Literature review

A literature review answers "what is already known" before you spend effort rediscovering it. Done well it maps a field's consensus, disagreements, and gaps; done badly it is a pile of citations proving nothing. The discipline is systematic search, honest quality assessment, and synthesis into a coherent picture.

Method

  1. Define the question and scope first. A specific question ("what methods reduce X under constraint Y") bounds the search; an unbounded "everything about the topic" drowns you. Set inclusion criteria (relevance, recency, quality bar) before searching, so you filter consistently rather than by whatever you find first (see research-synthesis).
  2. Search broadly, then systematically. Start with surveys and review papers (they map the field and cite the key work), then primary sources; use multiple channels (databases, citation search, domain venues: see citation-management, paper-lookup) because any single search misses things. Record your search terms and sources so the review is reproducible, not a lucky accident.
  3. Chase citations both directions. Backward (what a key paper cites: the foundations) and forward (what cites it: the developments and critiques): citation chasing from a few seminal papers surfaces the important work faster than keyword search alone, and reveals the conversation a field is having with itself.
  4. Assess quality, do not just collect. Not all sources are equal: venue and peer-review status, methodology rigor, sample size and reproducibility (see source-evaluation, fact-checking), citation count as a weak signal, and recency vs foundational-ness. Weight the strong evidence; note but discount the weak. A review that treats a blog post and a replicated study as equal is not a review.
  5. Synthesize into themes, not a list. Organize by idea (what is agreed, what is contested, what is unknown), not paper-by-paper: a matrix of sources against dimensions (approach, findings, limitations: see research-synthesis) reveals the structure. The synthesis is the value; an annotated bibliography that never connects the sources has done the collecting but not the review.
  6. Find the gap and the disagreements. The review's payoff is locating what is not yet known (your opportunity: see hypothesis-generation) and where the field disagrees (the live questions). Note the limitations the literature itself acknowledges; the honest gaps are where the useful work is.

Boundaries

  • A literature review establishes what is known and claimed; it does not verify the claims (that is replication and critical appraisal: see fact-checking). Consensus in the literature can be wrong, and a review should flag weakly-supported consensus.
  • Scope discipline is essential: the goal is enough coverage to ground the decision or research, not exhaustive coverage of an unbounded field. Know when you have enough (theme saturation: new sources stop adding new ideas).
  • For fast-moving fields, preprints and recent work matter more than citation count (which lags); for mature fields, foundational highly-cited work anchors. Adjust the recency-vs-authority weighting to the field.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.