literature-survey

Builds an evidence and contradiction matrix from iterative literature searches until coverage saturates.

Updated Sep 9, 2026
One-click install
npx skills add https://github.com/xlinh2301/EditCTC --skill literature-survey-xlinh2301
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: literature-survey
Source: https://github.com/xlinh2301/EditCTC/tree/main/.agents/skills/literature-survey
Command: npx skills add https://github.com/xlinh2301/EditCTC --skill literature-survey-xlinh2301

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? One-shot literature summaries miss contradictions, gaps, and coverage blind spots. This Skill runs a structured search-extract-map-expand loop that keeps pulling new sources until the evidence picture stops changing, producing a citation-backed matrix instead of a tidy but shallow narrative. ## Core Features & Use Cases - Evidence/Contradiction Matrix: Maps every source's stance (supports / contradicts / qualifies) on each claim, with verbatim snippets from real retrievals, validated against a JSON schema. - Saturation-Driven Loop: Expands queries and walks the citation graph each round, stopping only when new matrix-changing sources fall below a threshold for consecutive rounds or the round budget is hit. - Honest Synthesis: Outputs a survey organized into consensus, disputes, and gaps, plus a coverage note naming thin sub-topics. - Use Case: Ask for a survey of parameter-efficient fine-tuning methods since 2021; the Skill seeds sub-topic searches, admits sources round by round, records where papers disagree, and stops when rounds stop adding new evidence. ## Quick Start Ask the assistant to run a saturating literature survey on your research question and confirm the scope, depth preset, and round budget when prompted.

Frequently Asked Questions about literature-survey

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

FAQPage Schema
How do I run a structured literature survey with an evidence matrix?▼

State your research question with scope (years, sub-fields, inclusion criteria) and confirm the bindings: depth preset, round budget, patience, and saturation threshold. The loop then searches, extracts verbatim snippets, maps stances per claim, and stops at saturation or budget.

What is an evidence and contradiction matrix in a literature review?▼

It is a structured table of claims versus sources where each cell records a source's stance (supports, contradicts, or qualifies) backed by a verbatim snippet. Claims with both supporting and contradicting sources are flagged as contested.

How does the survey decide when to stop searching?▼

Stopping is measured, not guessed: a round is dry if it adds fewer than the configured minimum of new matrix-changing sources, and the loop halts after the patience count of consecutive dry rounds or when the round budget is reached.

Does the literature survey work without a Semantic Scholar API key?▼

Yes. A free S2_API_KEY improves snippet and citation reliability, but without it the toolchain degrades to a keyless pool and then to WebSearch/WebFetch, with that evidence tagged as web-sourced. The loop never blocks on a missing key.

When should I not use this survey loop?▼

Do not use it to grade a written proposal against the literature or to generate new hypotheses; it only maps what existing literature says. For a quick scan, run just the seed round instead of the full saturation loop.