Field-by-field comparison

Compare the evidence before choosing an MCP.

Each signal stays separate, missing facts remain visible, and the column order follows your selection.

MCPs in this comparison

Uncheck an MCP to remove it. Comparing two or three keeps each signal in its own row.

Open canonical comparison link

Comparable alternatives

Shown only when capability or industry context overlaps.

Side-by-side MCP evidence comparison
Evidence fieldCodebase Memory MCPMCP100 Recommendedio.github.blazickjp/arxiv-mcp-serverMCP100 Recommended
Task fit
Best for
  • Repository indexing, symbol search, call-path tracing, impact analysis, and architecture queries across many languages
  • Researchers who want focused arXiv literature workflows with local storage and bounded reads
  • Agents that need original LaTeX sections, citation graphs, BibTeX export, and topic watches without ingesting whole papers into context
Not ideal for
  • Tiny repositories, one-off text searches, or semantic questions unsupported by the indexed parsers
  • Users who need a fully hosted zero-local-state service
  • Workflows that cannot safely treat paper text and LaTeX as untrusted content
Avoid when
  • The source tree cannot be indexed locally or persistent cache files are prohibited
  • Your agent automatically acts on instructions embedded in retrieved papers without approval
  • You expose the optional HTTP transport publicly without upstream authentication and explicit Host/Origin controls
Provenance
Provenance details

First-party MCP

DeusData · Publisher source ↗

Community implementation

reviewed-promotion:io-github-blazickjp-arxiv-mcp-server · Publisher source ↗

Maintenance
Maintenance details

Repo: Aug 14, 2026

Package: Aug 14, 2026

Repo: Aug 26, 2026

Package: Not documented

Popularity evidence
GitHub stars

44,904

GitHub stars · checked 2026-09-25T18:35:00Z

3,173

GitHub stars · checked 2026-09-24T04:14:39.538000Z

Package downloads / 30 days

28,000

package downloads / 30 days · checked 2026-08-14T00:00:00.000Z

No reliable download data
External adoption evidence
  • npm: package downloads: 28,015Exact attributable signal.npm: package downloads · Rolling 30 days · checked 2026-08-14T18:46:26.000ZEvidence source ↗
Not documented
Client coverage
Client coverage details
Claude Code✓ Local stdio
Claude Desktop✓ Local stdio
Codex✓ Local stdio
VS Code✓ Local stdio
Cursor✓ Local stdio
OpenCode✓ Local stdio
Claude Code✓ Local stdio
Claude Desktop✓ Local stdio
Codex✓ Local stdio
VS Code✓ Local stdio
Cursor✓ Local stdio
OpenCode✓ Local stdio
Client coverage noteLocal stdio runs on your computer. Streamable HTTP connects to a remotely hosted MCP server.
Access and data
Authenticationlocal stdio; no remote auth neededDefault stdio operation requires no arXiv API key. citation_graph can optionally use SEMANTIC_SCHOLAR_API_KEY for higher reliability. If Streamable HTTP is enabled, the project recommends keeping it private or adding authentication upstream.
Costnone beyond local resourcesThe server is open source under Apache-2.0 and does not require an arXiv API key. External services may impose their own rate limits; an optional Semantic Scholar API key can improve citation reliability.
Permissionsreads indexed source/project data; installer writes supported agent/editor configs/hooksThe server searches external research services, downloads papers/source archives, and reads/writes its own local paper cache, topic watches, and semantic index under the configured storage path. It does not require broad account-level permissions.
Data handlinglocal processing; current docs state no source/index/query/env/telemetry uploadDownloaded papers, source archives, topic watches, and indexes are stored locally (default ~/.arxiv-mcp-server/papers). Search/source/citation/download requests go to external services. Paper text and LaTeX are explicitly treated as untrusted external input.
Limitations
Tradeoffs
  • Persistent indexing can reduce repeated reads, but consumes local storage and graph results can become stale until the repository is re-indexed.
  • Version 0.10.8 includes correctness and reliability fixes affecting graph/search results and Windows behavior.
  • Local storage and bounded reads improve privacy/context control, but downloaded papers, source archives, alerts, and indexes remain on the user's machine
  • Citation graphs depend on Semantic Scholar and optional API credentials; semantic search/reindex require the pro extra
Risk contextlatest 0.10.x supported; SLSA/Sigstore/checksums/VirusTotal/security policyLow to medium. Main risks are prompt injection in paper content, hostile LaTeX archives, local disk writes, and optional HTTP exposure. The project mitigates archive traversal/decompression risks, bounds content, binds HTTP to loopback by default, and enables DNS-rebinding protections.
Evidence date
Editorial review2026-09-232026-09-11
Candidate evidence2026-09-25T18:35:00Z2026-09-24T04:25:05.554391Z

Popularity, maintenance, fit, permissions, and client support are independent evidence fields.

Read the evidence method →