# plur [Health: Active]

**Category:** 🧠 Knowledge & Memory  
**Repository:** https://github.com/plur-ai/plur  
**GitHub Stars:** 308  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/plur

## Description
Open, local-first engram memory for AI agents: read, correct, delete; shared over MCP.

## Tools
Capabilities this server exposes over MCP:

- **plur_learn** — Store a correction, preference, or convention
- **plur_learn_batch** — Store many engrams in one call (batch dedup + per-item failure isolation)
- **plur_recall** — Retrieve relevant memories — hybrid (BM25 + embeddings) by default; `mode:"keyword"` for BM25-only
- **plur_inject_hybrid** — Select engrams for current task within token budget
- **plur_feedback** — Rate relevance (trains quality over time)
- **plur_forget** — Retire a memory (activation decays, eventually pruned)
- **plur_rescope** — Move an existing engram to another scope — personal → team, or back
- **plur_session_scope** — Change the session's default write scope mid-session
- **plur_capture** — Record an event — incident, resolution, session milestone
- **plur_timeline** — Query episode history by time, agent, or channel
- **plur_ingest** — Extract engrams from text automatically
- **plur_sync** — Sync via git. `personal` remotes mirror everything (use a private repo); `shared` remotes receive only shared-scope, non-private engrams
- **plur_status** — Check system health and engram counts
- **plur_receipt** — Counted, local report of what your memory retrieved for you
- **plur_outbox** — Inspect (and retry) team writes queued while their store was unreachable

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `uvx` (confidence: high):

```json
"mcpServers": {
  "plur": {
    "command": "uvx",
    "args": ["plur-hermes"]
  }
}
```

## Documentation

## What plur does

plur MCP server provides persistent memory for AI agents through plain-text engrams stored locally. An engram can represent a correction, preference, convention, incident, resolution, or other useful piece of operating knowledge. Agents can inspect, revise, retire, and retrieve these records rather than storing learned context only in a conversation or model weights.

The memory model supports personal, global, project, and team-oriented scopes. Scope is assigned per learned engram according to what the information concerns. Project-specific knowledge can be separated by domain, while global knowledge remains available across projects. A session's default write scope can also be changed during operation.

## How it works

Install the CLI and MCP package with Node.js 20 or newer, then run initialization and diagnostics. The initialization command configures the local store and the selected client. Diagnostics verify the setup and MCP handshake and can download the local search model when downloads are permitted.

Recall uses hybrid retrieval by default, combining BM25 keyword matching with embeddings. Passing keyword mode selects BM25-only retrieval. If the local search model is unavailable, search falls back to keyword retrieval. The system can also select memories for a task within a token budget, and relevance feedback can be used to improve retrieval quality over time.

Data stays local during normal operation. Git synchronization supports personal remotes that mirror everything and shared remotes that receive only shared-scope, non-private engrams. When a team store is unavailable, team writes can be queued and later reviewed or retried.

## Setup and configuration

Install both packages globally:

```bash
npm install -g @plur-ai/cli@latest @plur-ai/mcp@latest
plur init
plur doctor
```

Restart the editor after setup. The default local store is created under `~/.plur/`. For a multi-project Claude Code setup, initialization can specify a domain and scope:

```bash
plur init --domain myapp --scope project:my-app
```

Cursor setup uses `plur init --cursor`, which writes its MCP, hooks, and memory-rule files. Codex setup uses `plur init --codex` and registers the server in the user configuration. Client-specific configuration may expose a reduced Cursor tool profile; additional operations can be discovered through `plur_admin` with the `help` action.

## Tools and capabilities

- `plur_learn` stores one correction, preference, or convention.
- `plur_learn_batch` stores multiple engrams with deduplication and isolated per-item failures.
- `plur_recall` searches memories with hybrid or keyword retrieval.
- `plur_inject_hybrid` chooses relevant engrams within a token limit.
- `plur_feedback` records a relevance rating.
- `plur_forget` retires a memory for eventual pruning.
- `plur_rescope` moves an engram between scopes.
- `plur_session_scope` changes the session's default write scope.
- `plur_capture` records incidents, resolutions, or session milestones.
- `plur_timeline` filters episode history by time, agent, or channel.
- `plur_ingest` extracts engrams from text.
- `plur_sync` synchronizes memory through Git remotes.
- `plur_status` reports health and engram counts.
- `plur_receipt` produces a local counted report of retrieved memory.
- `plur_outbox` inspects and retries queued team writes.

_Full upstream README: https://allmcps.com/mcp/plur/readme_