litkg-rs

name	litkg-rs
description	Use when working on the standalone Rust literature KG toolkit, especially registry merge, arXiv download, TeX parsing, Semantic Scholar REST enrichment, graphify materialization, Neo4j export, and client-repo adapter configs.

litkg-rs

Grounding

Before changing the toolkit:

1. Read AGENTS.md
2. Read README.md
3. Read docs/architecture.md
4. Read .agents/AGENTS_INTERNAL_DB.md
5. Inspect .agents/issues.toml and .agents/todos.toml

Workflow

• Keep litkg-core repo-independent.
• Put client-repo assumptions in config files under examples/.
• Treat graph adapters as sinks over the same normalized paper model.
• Use the native Semantic Scholar REST client for paper search, batch metadata enrichment, recommendations, and author lookup; keep Asta MCP optional until its separate key flow is verified.
• Prefer deterministic output over aggressive parsing heuristics.
• For working-tree review, pull-request review, or autoresearch winner gating, use .agents/skills/code-review-litkg-rs/SKILL.md.
• For bounded, evidence-driven benchmark or overnight iteration loops, use .agents/skills/autoresearch-litkg-rs/SKILL.md.
• When a user explicitly asks to generalize an instruction, capture it into the smallest durable scaffold surface that fits:
  • root AGENTS.md for repo-wide policy
  • .agents/AGENTS_INTERNAL_DB.md for stable operating facts
  • .agents/issues.toml or .agents/todos.toml for tracked follow-up work
  • this skill for repeatable operator workflow
• Keep Apple Silicon considerations explicit when evaluating viewers, long-running agents, and local performance-sensitive tooling.

Validation

• cargo fmt --all
• cargo test
• make agents-db