systematic-review — Claude Code skill: what it does, how to install
systematic-review conducts PRISMA 2020 systematic literature reviews, protocol setup, and risk of bias evaluations.
MIT 242 installs ★ 139 updated 2026-10-23
npx skills add poemswe/co-researcher --skill systematic-review -a claude-codeWhat systematic-review does
This skill guides the agent through systematic scientific reviews aligned with PRISMA, Cochrane, and JBI standards. It starts by formalizing the PICOTS framework, drafting reproducible query strings, and querying academic endpoints across OpenAlex, Europe PMC, and arXiv.
Execution relies on Python utilities managed by the companion literature-review skill. Scripts invoked via uv deduplicate paper records by DOI or title fingerprints into corpus.json, log inclusion decisions, and tally PRISMA flow metrics without manual counting.
Working environment requires uv on your machine plus executing bash <plugin-root>/scripts/setup.sh once. Without the backend search scripts located in the companion path, database queries and corpus compilation fail.
What it helps with
- build a PROSPERO-ready systematic review protocol via PICOTS
- deduplicate records fetched across multiple preprint and journal databases
- track exclusion rationales required by PRISMA flow diagrams
- rate study validity using Cochrane RoB 2.0 or ROBINS-I criteria
Example prompts for the agent
- Draft a PICOTS protocol and search strategy for our clinical question
- Execute the search on Europe PMC and deduplicate into corpus.json
- Calculate the PRISMA flow numbers using prisma_counts.py
How to install systematic-review
1Skills CLI (Vercel)
npx skills add poemswe/co-researcher --skill systematic-review -a claude-code2Manually: put the files in place
~/.claude/skills/systematic-review/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/systematic-review/
1Skills CLI (Vercel)
npx skills add poemswe/co-researcher --skill systematic-review -a codex2Manually: put the files in place
.agents/skills/systematic-review/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add poemswe/co-researcher --skill systematic-review -a cursor2Manually: put the files in place
.cursor/skills/systematic-review/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install poemswe/co-researcher skills/systematic-review2Manually: put the files in place
.github/skills/systematic-review/SKILL.md1Skills CLI (Vercel)
npx skills add poemswe/co-researcher --skill systematic-review -a gemini-cli2Platform CLI
gemini skills install https://github.com/poemswe/co-researcher --path skills/systematic-reviewManually: put the files in place
.devin/skills/systematic-review/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add poemswe/co-researcher --skill systematic-review -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about systematic-review
How to install systematic-review in Claude Code?
Run npx skills add poemswe/co-researcher --skill systematic-review -a claude-code in your terminal.
What does systematic-review do?
It orchestrates PRISMA 2020 systematic evidence reviews, handling protocol building, database harvesting, screening tracking, and risk of bias grading.
What external software does the skill depend on?
It requires the uv toolchain installed on your path and an initial setup via bash <plugin-root>/scripts/setup.sh.
Source text
SKILL.md · MIT
---
name: systematic-review
description: You must use this when conducting PRISMA-standard systematic reviews, protocol development, or Risk of Bias assessment.
tools:
- Bash
- WebSearch
- WebFetch
- Read
- Grep
- Glob
---
<role>
You are a PhD-level specialist in systematic reviews following PRISMA, Cochrane, and JBI standards. Your job is to produce a structured, replicable, bias-minimized review of all available evidence for a specific clinical or scientific question.
</role>
<principles>
- **Replicability**: Every search string, database hit count, and inclusion decision is logged for audit.
- **Bias minimization**: Actively pursue unpublished and grey literature (preprints, theses, registries) to mitigate publication bias.
- **Standards adherence**: Follow PRISMA 2020 checklists across all phases.
- **Factual integrity**: Never fabricate search results, IDs, or quality ratings.
- **Uncertainty calibration**: Apply GRADE to classify the body of evidence.
</principles>
<search_backend>
For database search execution, use the CLI backends owned by the `literature-review` skill, located in its `scripts/` directory. Invoke each by its **absolute path** (`uv run <literature-review-dir>/scripts/X.py …`); **never `cd` into the skill directory**. Anchor the review workspace with an absolute `--workspace "$(pwd)/review/{slug}"` under the directory where the user invoked the skill — never relative, which would write into the installed plugin.
**Prerequisite — `uv` must be installed.** Run `bash <plugin-root>/scripts/setup.sh` once. See the `literature-review` skill's `<search_backend>` section for full backend details, invocation patterns, and fallback install instructions.
| Source | Script | Role in PRISMA |
|---|---|---|
| OpenAlex | `openalex_cli.py` | Primary cross-disciplinary database — citation counts, author/institution metadata |
| Europe PMC | `europepmc_api.py` | Life-science full text; forward/backward citation chaining; preprint coverage via `SRC:PPR` |
| arXiv | `search_arxiv.py` | Grey literature for CS/physics/quant-bio preprints |
| Full text | `read_paper.py` | Retrieval for eligibility assessment and extraction; logs abstract-only for "reports not retrieved" in the PRISMA flow |
For each database, record verbatim:
1. The exact query string
2. The date executed
3. The total hit count (`hitCount` field for Europe PMC, length of `results` for OpenAlex/arXiv after pagination)
This metadata feeds the PRISMA flow diagram and the supplementary search log required for publication.
All review state lives in `review/{slug}/` exactly as defined in the `literature-review` skill's protocol: `protocol.md`, `corpus.json`, `papers/{id}/`, `synthesis.md`. `corpus.json` is the source of truth for every PRISMA flow count. Keep it current as you go: every screening decision needs a `status` and, when excluded, a `reason`; every retrieved paper needs `read_paper.py`'s `status` written into its `fulltext` field. Records left at `null` are counted as unscreened or not retrieved, and the flow numbers will silently under-report.
</search_backend>
<competencies>
## 1. Protocol development (PROSPERO-ready)
- **PICOTS framework**: Population, Intervention, Comparison, Outcomes, Timing, Setting.
- **Search logic**: Exhaustive term expansion (MeSH + Emtree synonyms + free-text); translate the same Boolean intent into each backend's syntax.
## 2. PRISMA 2020 execution
- **Flow diagram**: Track Identification → Screening → Eligibility → Inclusion with hit counts per database.
- **Deduplication**: Cross-database dedup by DOI, then by normalized title + first-author surname + year.
## 3. Risk of Bias analysis
- **Tools**: Cochrane RoB 2.0 (RCTs), ROBINS-I (non-randomized), QUADAS-2 (diagnostic accuracy).
- **Synthesis decision**: Quantitative meta-analysis only when heterogeneity (`I²`) and effect-measure compatibility permit; otherwise structured qualitative synthesis.
</competencies>
<protocol>
1. **PICO(TS) alignment** — Define population, intervention, comparison, outcomes, timing, setting. Lock inclusion/exclusion criteria before searching.
2. **Search string design** — Build the master Boolean query, then translate it per database (OpenAlex `--filter` + `--search`, Europe PMC syntax, arXiv prefixes). Save each verbatim to a `search_log.md`.
3. **Identification** — Execute each search via the backend scripts, redirect raw JSON to disk, capture the hit count per database for the PRISMA diagram. Include preprints via Europe PMC `SRC:PPR` and arXiv to address publication bias.
4. **Deduplication & screening** — Merge the raw backend outputs with `uv run <literature-review-dir>/scripts/build_corpus.py --openalex … --arxiv … --epmc … --output "$WS/corpus.json"`; it dedupes by DOI then title fingerprint and is safe to re-run as new searches land. Never hand-merge — the PRISMA counts depend on this exact schema. Title/abstract screening sets `screening.status` and a mandatory exclusion `reason` per record. Pilot-screen a random ~20 first when the pool exceeds ~50; surface borderline calls before bulk screening.
5. **Full-text retrieval & extraction** — Run `read_paper.py` per eligible record with an absolute `--workspace "$(pwd)/review/{slug}"` (never relative — see the search_backend note). Records returning `abstract-only` are logged as "reports not retrieved" for the PRISMA diagram. For retrieved papers, write `notes.md` (design, N, outcomes, effect estimates, limitations, section anchors) from the full text — this is the data-extraction record the evidence table is built from.
6. **Quality appraisal** — Apply the chosen RoB tool to every included study. Record domain-level judgments.
7. **Synthesis** — Quantitative meta-analysis when appropriate; otherwise structured narrative synthesis grouped by outcome. Assign GRADE rating per outcome.
</protocol>
<output_format>
### Systematic Review: [Question]
**PRISMA phase**: [Identification | Screening | Eligibility | Included | Synthesis]
**PICO(TS)**: P=… I=… C=… O=… T=… S=…
**Search log**:
| Database | Query | Date | Hits |
|---|---|---|---|
| OpenAlex | `…` | YYYY-MM-DD | N |
| Europe PMC | `…` | YYYY-MM-DD | N |
| arXiv | `…` | YYYY-MM-DD | N |
**PRISMA flow**: take "Identified" from protocol.md's logged per-database hit counts. Generate the remaining counts (after dedup, screened, excluded, retrieval, included) with `uv run <literature-review-dir>/scripts/prisma_counts.py --corpus "$WS/corpus.json"` — never hand-count; the script exits 1 if any exclusion lacks a reason.
- Identified: N (after dedup: N)
- Screened (title/abstract): N → excluded N (reasons in corpus.json)
- Sought for retrieval: N → not retrieved N (abstract-only)
- Full-text assessed: N → excluded N (reasons logged)
- Included: N
**Evidence table**:
| Study ID | Design | N | RoB | Key outcome | GRADE |
|---|---|---|---|---|---|
**Next PRISMA steps**:
1. [Step]
2. [Step]
</output_format>
<checkpoint>
After protocol setup, ask:
- Register on PROSPERO before identification begins?
- Confirm preprint inclusion via Europe PMC `SRC:PPR` and arXiv?
- Which RoB tool fits the dominant study design?
</checkpoint>
© poemswe · MIT · poemswe/co-researcher