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 fieldGraphiti MCPShortlist candidateCodebase Memory MCPMCP100 Recommended
Task fit
Best for
  • Self-hosted persistent memory and knowledge-graph workflows
  • Teams willing to isolate the service behind network/auth controls
  • Repository indexing, symbol search, call-path tracing, impact analysis, and architecture queries across many languages
Not ideal for
  • Internet-exposed deployments without a reverse proxy or other access control
  • Production memory stores that require built-in strong authentication
  • Tiny repositories, one-off text searches, or semantic questions unsupported by the indexed parsers
Avoid when
  • You would expose the default 0.0.0.0 HTTP endpoint on an untrusted network
  • You cannot back up or recover graph data before enabling delete/clear tools
  • The source tree cannot be indexed locally or persistent cache files are prohibited
Provenance
Provenance details

First-party MCP

reviewed-detail:graphiti-mcp · Publisher source ↗

First-party MCP

DeusData · Publisher source ↗

Maintenance
Maintenance details

Repo: Not documented

Package: Not documented

Repo: Aug 14, 2026

Package: Aug 14, 2026

Popularity evidence
GitHub starsNot documented

44,904

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

Package downloads / 30 daysNo reliable download data

28,000

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

External adoption evidenceNot documented
  • npm: package downloads: 28,015Exact attributable signal.npm: package downloads · Rolling 30 days · checked 2026-08-14T18:46:26.000ZEvidence source ↗
Client coverage
Client coverage detailsNot documented
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
AuthenticationCurrent MCP server docs do not establish built-in HTTP authentication as a default control; deployments should rely on local access or external network/reverse-proxy controls.local stdio; no remote auth needed
CostOpen-source server; database and LLM/embedder provider costs apply.none beyond local resources
PermissionsTools include add/search plus delete_episode, delete_entity_edge, and clear_graph. Database privileges are those of the configured Neo4j/FalkorDB credentials.reads indexed source/project data; installer writes supported agent/editor configs/hooks
Data handlingStores knowledge-graph episodes/entities/facts and uses configured LLM/embedder providers during ingestion/search. Anonymous telemetry can be disabled.local processing; current docs state no source/index/query/env/telemetry upload
Limitations
Tradeoffs
  • Rich memory tools are valuable but include destructive operations
  • Multiple LLM/database credentials increase the deployment's secret surface
  • 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.
Risk contextHigh. Default network exposure plus destructive graph tools and backend/LLM credentials create material unauthorized-access and data-loss risk.latest 0.10.x supported; SLSA/Sigstore/checksums/VirusTotal/security policy
Evidence date
Editorial review2026-09-112026-09-23
Candidate evidence2026-09-11T00:00:00.000Z2026-09-25T18:35:00Z

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

Read the evidence method →