systematic-literature-retrieval

name	Systematic Literature Retrieval
description	Operations for federated API querying across ten literature engines (arXiv, Semantic Scholar, OpenAlex, Crossref, PubMed, SovietRxiv, ChinaRxiv, Europe PMC, bioRxiv, medRxiv).

Instructions

You are interacting with the src/literature/ module. This coordinates multi-source API harvesting across ten independently toggled engines, canonical-identifier deduplication, full-text resolution/extraction, unified bibliography export, and keyword-based subfield classification logic.

Agentic Interface (MCP Strategy)

1. Federated Politeness: Ensure rigorous rate-limiting across provider networks. If writing tests for this package, absolutely enforce the "Zero-Mock" policy by orchestrating local pytest-httpserver fixtures rather than blinding the agent via mocker.patch.
2. Canonical Deduplication: The ecosystem is messy; always prioritize cross-referencing DOIs, arXiv IDs, and fuzzy-hashed title matches.
3. Taxonomy Enforcement: Categorization logic into the A (Theory), B (Tools), C (Application) tiers relies heavily on specific mathematical token matching. Update the keyword heuristics with precision.