AlphaXiv

Academic paper search, reading, Q&A, code inspection, and annotations via the alpha CLI.

AlphaXiv is Feynman’s primary academic paper platform. It provides access to a large corpus of research papers along with search, full-text reading, targeted Q&A, linked code repository inspection, and persistent annotations. The Researcher agent uses the alpha CLI for all paper operations during research workflows.

Authentication

AlphaXiv requires authentication. Set it up once during initial setup or at any time:

feynman alpha login

Check your authentication status:

feynman alpha status

Log out:

feynman alpha logout

Authentication tokens are stored in ~/.feynman/auth/ and persist across sessions. No additional configuration is needed beyond logging in.

If you choose not to authenticate with AlphaXiv, Feynman falls back to web search for finding papers. This works for well-known work but misses citation metadata, discussion threads, and full-text access. For serious research workflows, AlphaXiv authentication is strongly recommended.

Commands

All paper operations use the alpha CLI via bash. The Feynman alpha-research skill teaches the agent to use these commands automatically.

Command	Description
alpha search "<query>"	Search papers. Uses semantic mode by default.
alpha search --mode keyword "<query>"	Exact-term keyword search.
alpha search --mode agentic "<query>"	Broader agentic retrieval.
alpha get <arxiv-id-or-url>	Fetch paper content and any local annotation.
alpha get --full-text <arxiv-id>	Get raw full text instead of an AI report.
alpha ask <arxiv-id> "<question>"	Ask a targeted question about a paper’s PDF.
alpha code <github-url> [path]	Read files from a paper’s linked GitHub repo. Use / for a repo overview.
alpha annotate <paper-id> "<note>"	Save a persistent annotation on a paper.
alpha annotate --clear <paper-id>	Remove an annotation.
alpha annotate --list	List all saved annotations.

Examples

# Search for papers on a topic
alpha search "transformer scaling laws"
alpha search --mode agentic "efficient attention mechanisms for long context"

# Fetch a paper by arXiv ID
alpha get 2106.09685

# Ask a targeted question about a paper
alpha ask 2106.09685 "What optimizer did they use?"

# Inspect code from a paper's repository
alpha code https://github.com/karpathy/nanoGPT src/model.py
alpha code https://github.com/karpathy/nanoGPT /

# Manage persistent annotations
alpha annotate 2106.09685 "Key paper on LoRA - revisit for adapter comparison"
alpha annotate --list
alpha annotate --clear 2106.09685

How the Researcher agent uses AlphaXiv

The Researcher uses the alpha CLI as its primary tool for academic content. Its selection logic:

Topic type	Tool
Academic papers, literature review	alpha search, alpha get, alpha ask
Current products, releases, pricing, docs	web_search, fetch_content
Mixed topics	Both, combined

For papers provided by arXiv ID, the Researcher fetches them directly with alpha get. For open-ended topics, it runs 2–4 varied alpha search queries simultaneously to map the landscape before narrowing.

When citing papers from alpha-backed tools, Feynman prefers direct arXiv or AlphaXiv links and includes the arXiv ID in the reference.

Use alpha search --mode agentic when broad retrieval is needed and --mode keyword only for exact-term lookups. Semantic mode (the default) performs best for most research queries.