skills.rest
wentorai
wentoraiCOMMUNITY
Explore all skills by this author →
GitHub

subagent-driven-review

Orchestrates parallel subagents to score, screen, and extract data from large literature searches.

850|114|Updated Mar 9, 2026
Search Tags:#subagents#systematic review#literature review#OpenAlex#scopus#citation traversal#parallel scoring
One-click install
npx skills add https://github.com/wentorai/Research-Claw --skill subagent-driven-review
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: subagent-driven-review
Source: https://github.com/wentorai/Research-Claw/tree/main/skills/subagent-driven-review
Command: npx skills add https://github.com/wentorai/Research-Claw --skill subagent-driven-review

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Screening 50+ papers from a literature search overwhelms a single context window and wastes time on serial scoring. This Skill coordinates parallel subagents to score abstracts, deep-dive full texts, and traverse citation networks while keeping the main context clean for consolidation.

Core Features & Use Cases

  • Parallel Relevance Scoring: Fetch abstracts once via rp_search, then fan out scoring subagents that read from the shared disk cache without triggering API rate limits.
  • Deep Dive Extraction: Dispatch up to 3-4 subagents running rp_fulltext to extract methods, measurements, and data availability from priority papers.
  • Citation Network Exploration: Run rp_cite per seed paper to collect backward and forward references, then score and dedupe into the review queue.
  • Use Case: A Scopus search returns 100 candidates. The main agent fetches abstracts once, splits records into 5 batches, dispatches 5 scoring subagents in parallel, and consolidates results into papers-reviewed.json sorted by relevance score.

Quick Start

Ask the agent to run a subagent-driven review: search Scopus for your query, then score the results in parallel batches and consolidate them into papers-reviewed.json.

Frequently Asked Questions about subagent-driven-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I screen 100 papers from a literature search with AI?▼

Fetch all abstracts once with rp_search, split the records into batches of 15-25, and dispatch parallel subagents to score each batch against a shared rubric. The main agent then merges, dedupes, and writes results to papers-reviewed.json.

Do parallel subagents speed up literature search API calls?▼

No. All subagents share the same IP and server-side rate limits, so API-bound steps like search and abstract fetching do not parallelize. Only reasoning-bound work like scoring and extraction speeds up; keep live-API subagents to 3-4 maximum.

When should I not use subagents for a literature review?▼

Avoid subagents for searches under 20 papers, since coordination overhead outweighs the benefit. Also skip them when papers require cross-comparison during screening or when you need real-time visibility into every decision.

Why do parallel subagents trigger 429 rate limit errors?▼

Each subagent runs its own per-process rate limiter, so they do not coordinate and can collectively exceed the per-IP ceiling. Pre-warm the shared disk cache first and cap live-API subagents at 3-4 so cached reads replace network calls.

How do subagents avoid conflicts when writing review results?▼

Subagents never write tracking files; they return JSON only. The main agent merges all responses, dedupes by DOI, sorts by score, and writes the single papers-reviewed.json file, eliminating write conflicts.

Related Skills

Explore More Education & Research Skills→
RenJW418RenJW418

foreign-literature-search

Search OpenAlex for social-science papers and generate Excel outputs with boolean query strings.

Community
Intermediate
kthornkthorn
118

Answering Research Questions

Automate literature reviews with search, evaluation, and synthesis phases.

Community
Basic
fagemxfagemx
2

self-check

Score Prismstack product skills against a 15D rubric and save results.

Community
Advanced
kthornkthorn
118

Subagent-Driven Literature Review

Orchestrate subagents for parallel literature screening and deep-dive extraction.

Community
Advanced