Distribution: submit PLUR to mcp.directory Claude Code memory roundup

[plur9]

Context

The mcp.directory roundup "Claude Code Memory MCP Servers (2026)" lists 4 memory MCP servers for Claude Code:

• Basic Memory
• Beads
• RAG Memory
• Knowledge Graph Memory

PLUR (@plur-ai/mcp) is not listed. This is a top discovery channel for our target audience (Claude Code users).

Action needed

Submit PLUR to mcp.directory for inclusion in their memory MCP roundup.

• If mcp.directory has a GitHub repo or submission form: submit directly
• If contact-only: email the site maintainer
• Key differentiators to highlight: engrams, knowledge packs, exchange protocol, local-first $0/query

Priority

P2 — not blocking but important for organic discovery. Do after #516 (ClawHub listing) since that is P0.

Related

• P0: Ship PLUR OpenClaw plugin to ClawHub — Supermemory beat us there #516 (P0 ClawHub listing — same distribution gap, higher priority)
• awesome-openclaw PR filed 2026-07-09: feat: add plur-ai/plur to Memory and Context Systems vincentkoc/awesome-openclaw#110

[plur9]

Research findings

PLUR is not listed — confirmed 404 at mcp.directory/servers/plur.

The Claude Code Memory MCP Servers (2026) roundup lists exactly 4 servers, none of which is PLUR:

Server	Differentiator
Basic Memory	Human-readable markdown files, fully editable
Beads (Steve Yegge)	Multi-day task graphs with dependency tracking
RAG Memory	Vector + knowledge graph hybrid retrieval
Knowledge Graph Memory	Anthropic's official minimal reference implementation

Two actions required

1. Submit to directory (2 min, form at mcp.directory/submit)

Field	Value
GitHub Repository URL	https://github.com/plur-ai/plur
npm Package	@plur-ai/mcp
Short Description	Engram-based memory MCP for Claude Code — structured recall, knowledge packs, hybrid search. Local-first, $0/query.
Email	gregor@ethswarm.org

2. Request roundup inclusion (contact form)

The roundup has no editorial email — use mcp.directory/contact. Draft:

Subject: PLUR (@plur-ai/mcp) — suggestion for Claude Code Memory roundup

Hi — I'm the author of PLUR, a memory MCP for Claude Code that doesn't appear in your Claude Code Memory MCP Servers (2026) roundup. Happy to share details if you're doing a refresh.

What's different: PLUR uses typed, decay-weighted "engrams" rather than raw notes or graph edges — each memory unit has relevance scoring and a session lifecycle (inject on start, score on end). The key unique feature is knowledge packs: installable, shareable bundles of domain engrams (e.g. a "TypeScript conventions" pack the whole team runs). We also have hybrid BM25 + embedding search and an exchange protocol for multi-agent memory sharing.

Repo: https://github.com/plur-ai/plur · npm: @plur-ai/mcp · Docs: https://plur.ai

PLUR's positioning vs. the 4 listed

• vs. Basic Memory: more structured (typed engrams vs. flat markdown) + searchable
• vs. Beads: session/recall focus vs. task-graph focus — different use case
• vs. RAG Memory: comparable retrieval quality + adds pack ecosystem
• vs. Knowledge Graph: deeper recall, lifecycle management, packs — at $0/query like all others

The pack ecosystem is the unique angle none of the 4 have. Lead with that in any outreach.

[plur9]

Submission ready — needs human to fill form

Submit URL: https://mcp.directory/submit

The form takes a GitHub repo URL and auto-pulls metadata within 24 hours. Fill in:

Field	Value
GitHub Repository URL	https://github.com/plur-ai/plur
npm Package URL	https://www.npmjs.com/package/@plur-ai/mcp
Short Description	see below
Email	gregor@ethswarm.org

Description (copy-paste):

Engram exchange layer — persistent memory for AI coding agents built on named, scored, domain-tagged memory primitives (engrams). Local-first: no cloud routing, no API key, data stays on-device. Hybrid BM25 + semantic recall. Knowledge packs let you share or sell curated memory sets. Best for power users who want memory that learns from corrections and survives across every tool in the stack.

Context

The existing roundup covers Basic Memory, Beads, RAG Memory, and Knowledge Graph Memory. PLUR's differentiator is the exchange layer: engrams as shareable primitives, not just personal notes. None of the four listed tools have a pack ecosystem or typed memory format.

🤖 heartbeat — 2026-07-09

[plur9]

Submission path confirmed — ready to go

mcp.directory/submit is a dead-simple form:

• Required: GitHub repository URL (auto-detects name, description, tools, README)
• Optional: npm package URL, short description, your email (for publication notification)
• Review within 24 hours, then auto-listed

Submission details (copy-paste ready)

GitHub URL:

https://github.com/plur-ai/plur

npm URL:

https://www.npmjs.com/package/@plur-ai/mcp

Short description (leads with exchange layer, per positioning — 2026-07-08 review):

Persistent memory for AI agents built on the engram exchange protocol. Store structured memories (engrams) with ACT-R decay, hybrid BM25+embedding recall, and knowledge pack sharing. Zero cloud dependency — memory stays local. Install: npx @plur-ai/mcp init

Action needed

Go to https://mcp.directory/submit, paste the URLs above, optionally paste the description, submit. ~30 seconds of human time.

Priority context: 4 competitors are listed (Basic Memory, Beads, RAG Memory, Knowledge Graph Memory) — this roundup is the top organic discovery surface for our target audience (Claude Code users searching for memory MCP servers).

🤖 heartbeat — ceo.distribution-check 2026-07-09

[crtahlin]

Status verification — partially done

Done:

• Research complete — confirmed PLUR is not listed on mcp.directory; compared against the 4 listed servers (Basic Memory, Beads, RAG Memory, Knowledge Graph Memory); differentiators identified (engram exchange layer, pack ecosystem)
• Full submission details prepared with copy-paste ready fields (GitHub URL, npm URL, short description, contact email)
• Roundup inclusion outreach draft written (2026-07-09 comment)

Still open:

• Actual form submission at mcp.directory/submit not yet completed — requires a human to fill the web form (~30 seconds); no confirmation of submission in issue history
• Roundup inclusion contact not yet sent

Leaving open.