get-research-paper
- Repo stars 0
- Author updated Live
- Author repo skills-registry
- Domain
- Writing
- Compatible agents
-
- Claude Code
- Cursor
- Cline
- Codex
- Windsurf
- Gemini CLI
- +20
- Trust score
- 88 / 100 · community maintained
- Author / version / license
- @tomevault-io · no license declared
- Token usage
- Lean
- Setup complexity
- Plug-and-play
- External API key
- Not required
- Operating systems
- Unspecified (assume cross-platform)
- Runtime requirements
- No special requirements
- Permissions
-
- Read-only
- Write / modify
- Network behavior
- External requests
- Install commands
- 26 variants
Profile is derived at build time from SKILL.md and install vectors. Subject to drift from author intent.
Heads up: 未限定 allowed-tools，默认拥有全部工具权限。
---
name: get-research-paper
description: Discovers, retrieves, ranks, and summarizes real existing research papers on any topic. Searches…
category: writing
runtime: no special runtime
---
# get-research-paper output preview
## PART A: Task fit
- Use case: Discovers, retrieves, ranks, and summarizes real existing research papers on any topic. Searches arXiv, Google Scholar, PubMed, Semantic Scholar, and reputable open repositories; returns a curated reading list with verified DOIs, key findings, and citation-ready metadata. Activates on slash commands (`/get-research-paper`, `/find-paper`, `/fetch-paper`, `/papers-on`, `/scholar`) and natural-language requests like "get research paper on …", "find papers about …", "what are the top papers on …". Hands off cleanly to the `research-paper` skill for paper writing. Runtime-neutral — works with Claude Code, OpenCode, Cursor, Cline, Codex, Aider, Amp, and 50+ agents. Use when this capability is needed..
- Inputs: target material, constraints, expected output, and acceptance criteria.
- Evidence boundary: follow “1. When to activate / Slash commands / Natural-language patterns” and do not present inference as author intent.
## PART B: Execution result
- **01** The card summarizes the use case; runtime output centers on “Discovers, retrieves, ranks, and summarizes real existing research papers on any topic. Searches arXiv, Google Scholar, PubMed, Semantic Scholar, and reputable open repositories; returns a curated reading list with verified DOIs, key findings, and citation-ready metadata. Activates on slash commands (`/get-research-paper`, `/find-paper`, `/fetch-paper`, `/papers-on`, `/scholar`) and natural-language requests like "get research paper on …", "find papers about …", "what are the top papers on …". Hands off cleanly to the `research-paper` skill for paper writing. Runtime-neutral — works with Claude Code, OpenCode, Cursor, Cline, Codex, Aider, Amp, and 50+ agents. Use when this capability is needed.”.
- **02** When the source has headings, the agent prioritizes “1. When to activate / Slash commands / Natural-language patterns” so the result follows the author’s structure.
- **03** Typical output includes task judgment, concrete steps, required commands or file edits, validation, and follow-up options.
- **04** Risk context follows the fingerprint: read files, write/modify files; may access external network resources; usually needs no extra API key.
## Running Rules
- read files, write/modify files; may access external network resources; usually needs no extra API key.
- Validate with a small sample before expanding scope.
- Return the result, validation criteria, and next iteration options. Choose the trigger
The source mentions slash commands such as `/get-research-paper`, `/find-paper`, `/find-papers`, `/fetch-paper`, `/papers-on`; use them first when your agent supports command triggers.
Provide inputs and boundaries
Name target files or source material, expected output, forbidden changes, and whether network or shell access is allowed. Permission fingerprint: read files, write/modify files.
Validate with a small run
Start with a small task and check whether the result follows “1. When to activate / Slash commands / Natural-language patterns”. Inspect diffs, logs, previews, or tests before expanding scope.
Review before delivery
Confirm the final output includes a concrete result, evidence, and next action. If it stays generic, tighten inputs, boundaries, and acceptance criteria.
---
name: get-research-paper
description: Discovers, retrieves, ranks, and summarizes real existing research papers on any topic. Searches…
category: writing
source: tomevault-io/skills-registry
---
# get-research-paper
## When to use
- Discovers, retrieves, ranks, and summarizes real existing research papers on any topic. Searches arXiv, Google Scholar…
- Use it when the task has clear inputs, repeatable steps, and validation criteria.
## What to provide
- Target material, scope, expected result, and forbidden changes.
- Whether network, commands, file writes, or external services are allowed.
## Execution rules
- Organize steps around “1. When to activate / Slash commands / Natural-language patterns” and keep inference separate from source facts.
- read files, write/modify files; may access external network resources; usually needs no extra API key.
- Validate with a small sample before expanding the task.
## Output requirements
- Return the deliverable, key evidence, validation method, and next action.
- Mark missing information as unknown; do not invent commands, platforms, or dependencies. Evidence boundary and execution chain
The author source anchors workflow facts; repository files anchor sources and commands; Fluxly only adds fit, limitations, and quality judgment.
skill "get-research-paper" {
input -> user goal + target files + boundaries + acceptance criteria
context -> 1. When to activate / Slash commands / Natural-language patterns
rules -> SKILL.md triggers / order / output contract
runtime -> no special runtime | read files, write/modify files | may access external network resources
guardrails -> usually needs no extra API key + small-sample validation + diff/log review
output -> copyable result + checklist + next iteration
}Get Research Paper
A research-discovery skill. Where the research-paper skill writes
papers, this skill finds them. Give it a topic, get a ranked,
de-duplicated reading list of real existing papers with verified DOIs,
key findings, and ready-to-cite metadata.
This file is the entry point. Heavier guidance (per-source strategies, ranking criteria, summarization prompts) lives in topic folders and is loaded on demand.
1. When to activate
Slash commands
| Command | What it does |
|---|---|
/get-research-paper <topic> |
Curated reading list (default 10 papers) |
/find-paper <topic> |
Alias for /get-research-paper |
/find-papers <topic> |
Alias for /get-research-paper |
/fetch-paper <topic> |
Alias for /get-research-paper |
/papers-on <topic> |
Alias for /get-research-paper |
/scholar <topic> |
Quick scholarly summary (5 papers, 2-line summaries) |
Common options:
--n <N>— number of papers (default 10).--years <range>— e.g.2020-2024,last-5,since-2018.--source <src>—arxiv,scholar,pubmed,semantic-scholar,all(default).--depth <quick|standard|deep>— summary detail.--style <harvard|apa|ieee|...>— pre-format the bibliography.--audience <academic|technical|general>— adjust summary register.--handoff— emit abibliography.yamlready for theresearch-paperskill.
Natural-language patterns
- "get research paper on / about / for [topic]"
- "find research papers on [topic]"
- "find papers on / about [topic]"
- "what are the top papers on [topic]"
- "show me research on [topic]"
- "fetch papers about [topic]"
- "list papers on [topic]"
- "literature on [topic]" (shorter than
/literature-review) - "scholar [topic]"
Negative activation
Do NOT activate for:
- Requests to write a paper (route to
research-paper). - Requests to review or critique a draft (route to
research-paper). - Casual questions ("what is X?") that don't need scholarly sources.
- Pure code / API documentation lookup.
2. Output contract
Every run produces, at minimum:
- Reading list — N papers with:
- Title (full)
- Authors (first 3 + "et al." if more)
- Year
- Venue / journal / preprint server
- DOI / arXiv ID / URL
- 2–4 sentence summary (problem → method → finding → significance)
- Relevance score (1–5) and quality score (per
citation_enginerubric) - Cite key (lowercase author_year_word) ready for use
- Field briefing (optional, default ON for
--depth deep) — a 1-paragraph synthesis of where the field is and what the dominant approaches are. bibliography.yaml— canonical-format file ready to drop into theresearch-paperskill.Known-gaps.mdblock — every paper that couldn't be verified is surfaced with severity and recommended fix.
See templates/reading-list.md, templates/paper-summary.md,
templates/briefing.md.
3. Core principles
- Anchor to TODAY's date FIRST. Before any search, determine
today's actual date (via
date -u +%Y-%m-%d, runtime context, or asking the user). Never default to training-cutoff dates. Year ranges like--years last-3are computed from today. Full protocol:instructions/freshness.md. - Real papers only. Never invent papers, DOIs, authors, or
findings. Use only sources the model can verify (or honestly mark
[UNVERIFIED — offline]). - De-duplicate aggressively. Same DOI / arXiv ID / first-author + year + title prefix → one entry.
- Rank by relevance and quality. A bad paper that mentions the topic is less useful than a great paper that's two clicks adjacent.
- Cite-ready by default. Every entry has cite_key + DOI + ready-to-use formatted citation.
- Triangulation. For load-bearing claims, prefer ≥ 2 independent sources. Note when a finding rests on a single source.
- Honest about limits. Without web tools, the model relies on training-data knowledge — flag every entry accordingly.
- Hand off cleanly. Output is consumable by the
research-paperskill via--handoffmode.
4. Top-level workflow
intake → search-strategy → fan-out search → rank+dedupe →
verify → summarize → assemble briefing → output (+ optional handoff)Each step has a dedicated playbook. Read the file for the step you're
on; persist the artifact; move on. Master pipeline:
workflows/search.md.
5. Source coverage
| Source | When to prefer | Tool |
|---|---|---|
| arXiv | CS, ML, AI, physics, math, quant-bio | toolchains/arxiv_search.py (works offline-only via API) |
| Google Scholar | Generic / cross-discipline broad surveys | WebSearch with site:scholar.google.com |
| Semantic Scholar | API-friendly, citation graph, summaries | WebFetch of api.semanticscholar.org |
| PubMed / PubMed Central | Biomedical, life sciences | WebFetch of eutils.ncbi.nlm.nih.gov |
| DBLP | CS authors / venues / publication lists | WebFetch of dblp.org |
| ACM DL | HCI, systems, security, networks | WebSearch with site:dl.acm.org |
| IEEE Xplore | Engineering, signal, hardware | WebSearch with site:ieeexplore.ieee.org |
| OpenReview | NeurIPS, ICLR, ICML reviews + papers | WebFetch of openreview.net |
| Crossref | DOI verification + metadata fill-in | WebFetch of api.crossref.org |
| Retraction Watch | Retraction screening | WebFetch of retractionwatch.com / database |
Per-source strategy details: sources/.
6. Ranking and quality
Each candidate paper is scored on:
- Authority (0–4) — venue quality (peer-review rigor, impact).
- Methodological rigor (0–3) — replicability, sample size, sound stats.
- Recency / relevance (0–3) — fresh + topical, OR foundational + canonical.
- Total (0–10) — used to rank.
Default reading lists keep papers scoring ≥ 5. Higher floors raise
the bar (--quality-floor 7).
Full rubric: prompts/ranking.md (extends the
citation_engine/source-evaluation.md of the research-paper skill).
7. Handoff to research-paper
After producing a reading list:
/get-research-paper "graph neural networks for fraud detection" \
--n 25 --handoff --style ieee --years 2020-2024Produces:
gnn-fraud-detection/
├── reading-list.md # human-readable curated list
├── bibliography.yaml # ← canonical file for research-paper skill
├── briefing.md # 1-paragraph synthesis
└── Known-gaps.md # any unverifiable itemsThe user then runs the writer skill with the produced bibliography:
/research "graph neural networks for fraud detection" \
--style ieee --bibliography ./gnn-fraud-detection/bibliography.yamlThe writer reads the curated bibliography directly — no re-search needed.
8. Failure handling
- No web search available → use model-known papers, mark every
entry
[UNVERIFIED — offline], lower the recommended--nto 5–8, and surface the limitation in the briefing. - Search returns nothing → broaden the query (drop adjectives, try synonyms), then return what was found with an honest note.
- Conflicting metadata across sources → prefer the published (peer-reviewed) version over the preprint; note the relationship.
- Retracted paper detected → drop from the list; flag in
Known-gaps.md. - Out-of-scope topic → surface a note in the briefing; deliver best-effort results.
9. Where to look next
- Plan a search →
workflows/search.md - Per-source strategy →
sources/ - Ranking rubric →
prompts/ranking.md - Summarization →
prompts/summarization.md - Output templates →
templates/ - Hand off to writer →
workflows/handoff-to-writer.md - arXiv search tool →
toolchains/arxiv_search.py
This skill is intentionally smaller than the writer skill. Its job is
discovery and curation; the heavy lifting (writing, methodology,
review) lives in research-paper.
Source: aniketkrs/research-paper — distributed by TomeVault.
Decide Fit First
Design Intent
How To Use It
Boundaries And Review