RCLL (fleet-memory) — Watchlist

MCP100 Assessment

Watchlist / Tracking

Why we’re tracking it

Follow-up value is specific to RCLL hierarchical room-scoped shared memory: Does the Hindsight-derived fork preserve room boundaries and attribution across memory-depth changes?

Reviewed Oct 5, 2026 · Documentation review; complete setup workflow not independently tested

Best fit

Self-hosted shared memory for multiple AI agents using rooms, layers and local retrieval.

What we know

RCLL is a self-hosted shared-memory system backed by Postgres/pgvector. Reads use local embedding/reranking and do not invoke an LLM; writes can use no model, a local OpenAI-compatible model or a hosted compatible endpoint for fact extraction.

Maintenance / activity evidence

fleet-memory-mcp 0.1.0 published 2026-08-28T12:34:56.051Z

Tracking evidence observed Oct 5, 2026. These dated signals do not establish current maintenance. Reviewed enrichment confidence: high.

Verification gaps / watch-outs

Installation and comprehensive security have not been independently audited.
Shared-memory read/write authority across agent fleet. Writes: Retain/write agent facts and documents; Consolidate/modify memory state.
Deployment-dependent; may be keyless locally or use model-provider credentials.
Sensitive access: Cross-agent memory, Operational decisions, Potentially sensitive source/business context

Shared memory has broad confidentiality impact; self-hosting helps but write-path provider choice matters.

Full authority, access and evidence limits are available in Review context and Access & risk.

All limitations / Review context

Does the Hindsight-derived fork preserve room boundaries and attribution across memory-depth changes?

The bounded review of the publisher's documented RCLL hierarchical room-scoped shared memory and release notices identified no disclosed material condition preventing an accurately scoped public tracking entry. The resource's operational permissions remain uncertainty; this assessment does not certify installation safety.

Shared memory has broad confidentiality impact; self-hosting helps but write-path provider choice matters.
Shared-memory read/write authority across agent fleet.
Deployment-dependent; may be keyless locally or use model-provider credentials.
Memory is stored in operator-controlled Postgres. Read path is local by design; hosted write extraction sends document content to the configured compatible endpoint.
Reviewed write actions: Retain/write agent facts and documents; Consolidate/modify memory state
Sensitive access: Cross-agent memory, Operational decisions, Potentially sensitive source/business context
Installation, client compatibility and comprehensive security have not been independently audited. Authentication and access depend on the configured services.
Runtime behavior, dependencies and undisclosed vulnerabilities were not independently audited.
Documented tool authority and any remote data handling must be checked before installation or use.
Tool inventory, client matrix and permissions completeness were not requirements for this Radar tracking review.
Room assignment is currently immutable after write.
A wrongly scoped room can return zero relevant facts.
Export is incomplete for full fact/link portability.
Write-path fact extraction may require a model and can be slow locally.
The project documents retrieval limitations rather than claiming answer accuracy.
Project openly documents unresolved retrieval/export limitations; preserve them rather than presenting RCLL as a solved memory layer.

Original activity observations

fleet-memory-mcp 0.1.0 published 2026-08-28T12:34:56.051Z

Technical identity

npm/fleet-memory-mcp

Exact implementation source: https://www.npmjs.com/package/fleet-memory-mcp ↗
Reviewed resource type: MCP Server
Repository identity: Holetron-lab/fleet-memory
Package identity: npm:fleet-memory-mcp

Evidence & Sources

Evidence basis: Recorded source evidence
Setup runtime verification: Not independently runtime-tested by MCP100 for this setup workflow.

Reviewed Oct 5, 2026 · Latest source observation Oct 8, 2026

Package registry metadata ↗
fleet-memory-mcp
Recorded package identity, distribution or download evidence.
Evidence source ↗
rcll.ai/
Source material retained for this review.
Includes references whose attribution is not established.
GitHub repository ↗
holetron-lab/fleet-memory
Repository reference; implementation attribution is not established for every document.
Includes references whose attribution is not established.

Setup · Watchlist

Install & Agents

Publisher instructions.

Source evidence observed Oct 14, 2026.

Additional setup details required — see the specific gaps below.

Warning

Not runtime tested, no third-party installation, network authorization, paid calls, or mutation executed.
Do not claim production service is available based on README alone.
Publisher MCP configuration; deprecated env vars need explicit migration, inspect memory storage scope
Do not replace deprecated old memory environment variables with assumed new names; refer to package documented migration table.
Official npm publisher package page README inspected for this exact npm identity; static research only. Some setup statements are conservatively summarized; no source code or live client execution checked. V2 review did not run any live client.
Check actual host syntax and supply credentials in secure client environment; do not paste placeholders verbatim.
Avoid write, payment, token transfer, publishing, sending, booking and remote account changes as tests.
Replace credential, action and registration placeholders; follow the matching publisher/client prerequisites. No execution performed.
Use the matching host, actual registration name and original installation environment; this does not revoke credentials.

Install / connect
Publisher instructions
Publisher MCP configuration; deprecated env vars need explicit migration, inspect memory storage scope | Publisher npm README revisited on 2026-10-14; exact identity retained.
Reviewed setup example · Publisher instructions
npx fleet-memory-mcp
Configure / authenticate
MCP100 setup guide
Publisher-documented connection/installation path shown under install.commands. Check actual host syntax and supply credentials in secure client environment; do not paste placeholders verbatim. Publisher MCP configuration; deprecated env vars need explicit migration, inspect memory storage scope Verify the correct host/transport from the publisher README; secure credentials in local client settings.
Check connection
Suggested check
Check that the registered MCP tools are discoverable in the client; run only publisher-described read-only status/health tools if provided. No runtime verification was performed.
Remove / uninstall / revoke
MCP100 setup guide
If your Claude Code MCP registration is named `fleet-memory-mcp`, run the client removal command. For Claude Desktop/Cursor/Codex or another registered name, remove the matching entry in that host’s MCP settings. Hosted OAuth disconnect and stopping a self-hosted process are separate; npx has no permanent global package install.
Remove named client registration · Claude Code only · MCP100 composed guide
claude mcp remove fleet-memory-mcp

Official Documentation & Sources

What it does

RCLL is a self-hosted shared-memory system backed by Postgres/pgvector. Reads use local embedding/reranking and do not invoke an LLM; writes can use no model, a local OpenAI-compatible model or a hosted compatible endpoint for fact extraction.

Capability: Files & Knowledge
Resource type: MCP Server
Best fit: Self-hosted shared memory for multiple AI agents using rooms, layers and local retrieval.

Representative uses

Recall shared agent facts by room/layer
Retain documents/facts into shared memory
Share operational decisions across agents
Run completely local read path

Limitations

Room assignment is currently immutable after write.
A wrongly scoped room can return zero relevant facts.
Export is incomplete for full fact/link portability.
Write-path fact extraction may require a model and can be slow locally.
The project documents retrieval limitations rather than claiming answer accuracy.
Project openly documents unresolved retrieval/export limitations; preserve them rather than presenting RCLL as a solved memory layer.

Reviewed access

Access & risk

Review the exact access boundary before connecting this MCP to accounts, credentials, or sensitive data.

Authentication: Deployment-dependent; may be keyless locally or use model-provider credentials.
Permissions: Shared-memory read/write authority across agent fleet.
Data handling: Memory is stored in operator-controlled Postgres. Read path is local by design; hosted write extraction sends document content to the configured compatible endpoint.
Cost: self_hosted
Risk and cautions: Shared memory has broad confidentiality impact; self-hosting helps but write-path provider choice matters.

Access scope and cautions

Write actions: Retain/write agent facts and documents Consolidate/modify memory state
Sensitive access: Cross-agent memory Operational decisions Potentially sensitive source/business context
Risk level: high

FAQ

Based on reviewed evidence as of 2026-10-08. Watchlist tracking is not a recommendation or a security certification.

Which tasks are documented for this implementation?

The reviewed tasks include: Recall shared agent facts by room/layer; Retain documents/facts into shared memory; Share operational decisions across agents; Run completely local read path.

What setup and authentication are documented?

Recorded installation/entrypoint step: npx fleet-memory-mcp. Recorded transport: stdio, self_hosted. Recorded environment-variable notes: LLM_PROVIDER (optional); LLM_BASE_URL/LLM_MODEL for local or compatible provider; Postgres/pgvector configuration. No configuration values are supplied here. Prerequisites: Postgres with pgvector; Docker/self-hosted environment. These are the documented setup details, not a claim that installation or client compatibility was independently tested. Authentication: Read-only deployments can require no API key; write extraction may use local or hosted model credentials depending on configuration.

Can it modify data or perform write operations?

Documented write actions include: Retain/write agent facts and documents; Consolidate/modify memory state. Recorded permission boundary: Shared-memory read/write authority across agent fleet.

What data-handling boundary should I consider?

Memory is stored in operator-controlled Postgres. Read path is local by design; hosted write extraction sends document content to the configured compatible endpoint. Recorded sensitive-access context: Cross-agent memory; Operational decisions; Potentially sensitive source/business context.

What limitations and risks should I check before use?

Shared memory has broad confidentiality impact; self-hosting helps but write-path provider choice matters. Room assignment is currently immutable after write. A wrongly scoped room can return zero relevant facts. Export is incomplete for full fact/link portability. Write-path fact extraction may require a model and can be slow locally. The project documents retrieval limitations rather than claiming answer accuracy. Project openly documents unresolved retrieval/export limitations; preserve them rather than presenting RCLL as a solved memory layer. Runtime behavior, dependencies and undisclosed vulnerabilities were not independently audited. Documented tool authority and any remote data handling must be checked before installation or use.