
Related Work Writer
- 41 installs
- 1 repo stars
- Updated July 31, 2026
- jurgendn/agent-skills
Helps with ai & agent building tasks.
About
related-work-writer is a Claude Code skill for ai & agent building. It helps solo builders move faster with AI-assisted development.
- related-work-writer
- AI & Agent Building
- AI-coding skill
Related Work Writer by the numbers
- 41 all-time installs (skills.sh)
- Ranked #8,144 of 16,546 AI & Agent Building skills by installs in the Skillselion catalog
- Data as of Aug 2, 2026 (Skillselion catalog sync)
npx skills add https://github.com/jurgendn/agent-skills --skill related-work-writerAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| Installs | 41 |
|---|---|
| repo stars | ★ 1 |
| Last updated | July 31, 2026 |
| Repository | jurgendn/agent-skills ↗ |
What it does
Helps with ai & agent building tasks.
Files
Related Work Writer
Related work is not a bibliography in paragraph form. It is the part of the paper that teaches the reader which intellectual neighborhood the contribution belongs to and why the current paper is still needed.
Use this when
- The user has papers, notes, or a source map and wants Related Work prose.
- A draft reads like a citation laundry list.
- The user needs to position their work without overstating novelty.
- The user wants to group prior work thematically.
- The paper needs fair contrasts against baselines, methods, datasets, or theory lines.
Do not use this when
- The literature has not been gathered or checked. Use
literature-triangulationfirst. - The user needs the whole paper argument. Use
paper-argument-planner. - The user only needs BibTeX formatting or citation cleanup.
Workflow
1. Inventory the source material
For each source or group of sources, identify:
- What problem it addresses.
- What method, theory, benchmark, or finding it introduced.
- How it relates to the user's paper.
- What limitation or open question remains.
- Whether the contrast is supported by the source or inferred.
If a claim about prior work is not grounded in a provided or verifiable source, mark it as needing citation support.
2. Choose the organizing principle
Pick the structure that best supports the paper's argument:
- By problem lineage — when the field evolved through successive problem formulations.
- By method family — when techniques are the main contrast.
- By evidence/benchmark — when evaluation practice is the key issue.
- By theory vs empirical work — when the paper bridges or separates them.
- By limitation — when the current paper targets a specific gap.
Avoid chronological ordering unless the chronology itself explains the gap.
3. Write contrast fairly
Use calibrated language:
- Prefer: “Closest to our work is X, which studies Y; our setting differs in Z.”
- Prefer: “Prior work establishes A, but leaves B unresolved.”
- Avoid: “No prior work has considered...” unless the evidence is very strong.
- Avoid: dismissive contrasts that make prior work sound foolish.
The goal is to make the contribution look necessary, not to make prior work look bad.
4. Control citation density
Each paragraph should have a conceptual job. Citations support that job.
Good paragraph pattern:
1. Topic sentence naming the line of work. 2. Representative citations grouped by role. 3. What that line establishes. 4. How the current paper differs or builds on it.
Do not attach a citation to every sentence if one grouped citation set can support the paragraph.
5. Preserve uncertainty
If the source map has disagreements, say so. If the boundary between this work and prior work is subtle, explain the subtlety. Reviewers trust precise distinctions more than loud novelty claims.
Output format
When drafting a section, return:
# Related Work Draft
[Draft prose]
## Structure rationale
## Citation gaps or claims needing verification
## Novelty-risk notesWhen auditing an existing section, return:
# Related Work Audit
## What works
## Citation laundry-list issues
## Unfair or overbroad contrasts
## Missing lines of work
## Suggested rewrite planQuality bar
A good Related Work section lets a skeptical reviewer say: “They understand the field, they represented my area fairly, and the distinction they claim is specific enough to evaluate.”
{
"skill_name": "related-work-writer",
"evals": [
{
"id": 1,
"prompt": "I have notes on three groups of papers for a draft: retrieval-augmented QA systems, biomedical entity linking, and contrastive representation learning. Write a related work section that positions my method as terminology-aware retrieval training without claiming nobody has done biomedical retrieval before.",
"expected_output": "A related work draft grouped by conceptual role, with fair contrasts and citation-gap notes.",
"files": []
},
{
"id": 2,
"prompt": "This related work paragraph is just a list of citations. Help me reorganize it thematically and explain how to contrast our approach with prior prompt optimization and RAG evaluation papers.",
"expected_output": "A related work audit and rewrite plan that avoids citation laundry-list structure.",
"files": []
},
{
"id": 3,
"prompt": "Given a literature map with columns: paper, method family, benchmark, limitation, and relevance to us, turn it into a Related Work outline with paragraph jobs and novelty-risk warnings.",
"expected_output": "A source-grounded related work outline with paragraph roles and calibrated novelty cautions.",
"files": []
}
]
}