Compartment — Watchlist

MCP100 Assessment

Watchlist / Tracking

Why we’re tracking it

Follow encryption, local-only storage, and client adapters as the vault evolves

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

Best fit

Encrypted offline persistent memory shared across local agent clients.

What we know

Offline encrypted memory vault with hybrid search, per-agent namespaces, auditability and a stdio MCP server.

Maintenance / activity evidence

Observed Sep 30, 2026 (UTC)Tests: the refresh test gives Hermes its own home on Windows too

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

Activity source · commits ↗Activity source · releases ↗

Verification gaps / watch-outs

Installation and comprehensive security have not been independently audited.
Namespace rw/ro/none plus local process access. Writes: Store/forget/link memories; Import/export/install integrations.
Vault passphrase.
Sensitive access: Durable memories, Potential credentials stored as memory

Network exposure is low, but persistent memory can contain highly sensitive information.

Security source · repository notice ↗Security source · policy file ↗

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

Popularity

581 repository stars

Stars observed Sep 25, 2026.

Do not overwrite existing reviewed stars/download metrics with nulls from this enrichment batch.

All limitations / Review context

Which clients and vault operations remain supported without a hosted dependency?

The bounded assessment reviewed the official README, repository notices, security policy path, and security page for known material public-tracking blockers. No such blocker was identified. This is a public-tracking assessment, not an installation safety certification.

Network exposure is low, but persistent memory can contain highly sensitive information.
Namespace rw/ro/none plus local process access.
Vault passphrase.
Memories and embeddings are encrypted at rest; project states no runtime network/telemetry.
Reviewed write actions: Store/forget/link memories; Import/export/install integrations
Sensitive access: Durable memories, Potential credentials stored as memory
Installation, client compatibility and comprehensive security have not been independently audited. Authentication and access depend on the configured services.
No independent runtime, installation, or comprehensive security audit was performed.
No installation, runtime, client matrix, or complete tool-contract audit was performed in this review.
Python 3.11+ for pip route.
Plaintext export is intentionally unencrypted.
Optional memory-unlock tool is disabled by default because passphrases would enter model context.

Original activity observations

2026-09-30T09:06:05Z: Tests: the refresh test gives Hermes its own home on Windows too

Technical identity

github.com/maxfreedompollard/compartment

Exact implementation source: https://github.com/maxfreedompollard/compartment ↗
Reviewed resource type: MCP Server
Repository identity: maxfreedompollard/compartment
Package identity: pypi:compartment

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 7, 2026

GitHub repository ↗
maxfreedompollard/compartment
Recorded repository material and project history.
Includes references whose attribution is not established.

Setup · Watchlist

Install & Agents

claude-code

Publisher instructions with a documented client setup path.

Source evidence observed Oct 14, 2026.

Additional setup details required — see the specific gaps below.

Install / connect
Publisher instructions
Compartment is local encrypted agent memory with a user-initialized vault and optional client integrations.
package install · Publisher instructions
pip install compartment
self hosted install or run · Publisher instructions
compartment init
client registration · Publisher instructions
compartment integrate claude
Configure / authenticate
Publisher instructions
Compartment is local, keyless memory MCP. Install pip install compartment then compartment init; register stdio `compartment --caller <host> serve` using host-specific instructions.
Check connection
Suggested check
Check local MCP tool discovery; no external key revocation applies to default mode.
Remove / uninstall / revoke
MCP100 setup guide
Before removal, unregister Compartment from each MCP host and undo any optional hooks using its published CLI setup/help. Keep persistent memory unless explicitly deleted.
Distinguish client registration removal from software uninstall, persistent data deletion and provider OAuth/API-key revocation.
pip uninstall compartment

Official Documentation & Sources

What it does

Offline encrypted memory vault with hybrid search, per-agent namespaces, auditability and a stdio MCP server.

Capability: Files & Knowledge
Resource type: MCP Server
Best fit: Encrypted offline persistent memory shared across local agent clients.

Representative uses

Store/retrieve durable memories
Search with vector + keyword retrieval
Share one vault across clients
Use namespace-level read/write controls

Limitations

Python 3.11+ for pip route.
Plaintext export is intentionally unencrypted.
Optional memory-unlock tool is disabled by default because passphrases would enter model context.

Reviewed access

Access & risk

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

Authentication: Vault passphrase.
Permissions: Namespace rw/ro/none plus local process access.
Data handling: Memories and embeddings are encrypted at rest; project states no runtime network/telemetry.
Cost: free
Risk and cautions: Network exposure is low, but persistent memory can contain highly sensitive information.

Access scope and cautions

Write actions: Store/forget/link memories Import/export/install integrations
Sensitive access: Durable memories Potential credentials stored as memory
Risk level: moderate

FAQ

Based on reviewed evidence as of 2026-10-07. Watchlist tracking is not a recommendation or a security certification.

Which tasks are documented for this implementation?

The reviewed tasks include: Store/retrieve durable memories; Search with vector + keyword retrieval; Share one vault across clients; Use namespace-level read/write controls.

What setup and authentication are documented?

Recorded installation/entrypoint step: pip install compartment && compartment init. Recorded transport: stdio. Prerequisites: Python 3.11+. These are the documented setup details, not a claim that installation or client compatibility was independently tested. Authentication: Local passphrase unlocks encrypted vault.

Can it modify data or perform write operations?

Documented write actions include: Store/forget/link memories; Import/export/install integrations. Recorded permission boundary: Namespace rw/ro/none plus local process access.

What data-handling boundary should I consider?

Memories and embeddings are encrypted at rest; project states no runtime network/telemetry. Recorded sensitive-access context: Durable memories; Potential credentials stored as memory.

What limitations and risks should I check before use?

Network exposure is low, but persistent memory can contain highly sensitive information. Python 3.11+ for pip route. Plaintext export is intentionally unencrypted. Optional memory-unlock tool is disabled by default because passphrases would enter model context. No independent runtime, installation, or comprehensive security audit was performed.

Is this an official MCP?

MaxFreedomPollard is listed as First-party MCP in MCP100’s current evidence.