Exomem

MCP文档知识

提供方 未标注

文档知识 未标注 未标注 官方维护 Claude DesktopClaudeCursorVS Code

本地Markdown/Obsidian知识基底,为MCP代理提供受控记忆与混合搜索。

一句话选型结论

按资料字段整理

Exomem 属于文档知识类 MCP 服务器,传输方式未标注,鉴权方式未标注,可直接接入 Claude Desktop、Claude、Cursor 等,没有现成安装命令,需按官方文档接入。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类文档知识 传输方式未标注 鉴权未标注 部署方式未标注
GitHub 星标 未标注 仓库热度,无公开仓库时显示未标注
工具数 未标注 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 官方维护 核验等级 仅资料核对

工具与能力

文档知识
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量未标注暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

支持客户端

适用场景

记忆

安装与配置

未标注 未标注
上游未给出传输方式
  1. 1
    准备运行环境

    按官方文档确认运行环境(常见为 Node.js 或 Python),装好后再执行安装命令。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后随便调一个只读工具,确认能返回结果,再接入正式流程。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
文档检索问答

在你的文档库里按问题找答案。

知识库读写

把新资料写进知识库,随用随取。

内容整理

把零散资料归纳成结构化笔记。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 已适配 Claude Desktop、Claude、Cursor,换客户端不用重新研究配置。

先想清楚的地方

  • 没有现成安装命令,需要按官方文档自己填启动命令和参数。
  • 资料没标注工具数量,实际能力要装好后确认。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在文档知识场景里让模型直接调用外部能力的人。
  • 已经在用 Claude Desktop、Claude 的用户,接入路径最短。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
Exomem当前 未标注 未标注 未标注 无 未标注
dify 未标注 stdio · 本地进程 本地部署 有 2026-04-11
open-webui 未标注 stdio · 本地进程 本地部署 有 2026-04-11
JeecgBoot 未标注 stdio · 本地进程 本地部署 有 2026-04-11
Cherry Studio 未标注 未标注 未标注 无 2026-05-17
mempalace 未标注 stdio · 本地进程 本地部署 有 2026-04-11

服务器档案

文档知识
字段来自项目仓库与公开索引,中文说明由本站标注
提供方未标注上游标注的维护方 未标注
分类文档知识本站按上游分类展示
连接方式未标注上游未给出传输方式
鉴权方式未标注远程服务建议用最小权限凭据
维护方官方维护核验等级 仅资料核对
包名 / 运行时未标注 · 未标注本地安装时用到的包标识与运行时
部署类型未标注上游给出的部署形态
最近核验未标注上游最后核验时间

详细介绍

25.9 KB
来自项目仓库与上游目录的原文,未做删改

exomem

Durable memory with sources, proof, history, and review for MCP-capable agents.

exomem turns an owned Markdown/Obsidian vault into a local knowledge substrate
for Codex, Claude Code, Cursor, chatbots, CLI agents, and any client that can
call MCP tools. Your files stay plain, local, portable, and editable outside the
server.```text
agent -> MCP tools -> exomem -> your Markdown / Obsidian vault


## Prove it in 30 seconds```bash
uvx exomem demo

One command, no install, no config, no vault of your own needed:```text
exomem demo — bundled sample vault, keyword mode, fully local
vault: /tmp/exomem-demo-XXXXXX

  1. doctor: PASS (0.8s)
  2. find "retrieval": PASS (0.1s)
    - Knowledge Base/Sources/Sessions/2026-06-30-sample-session.md
    - Knowledge Base/Notes/Insights/retrieval-needs-owned-files.md
  3. get retrieval insight: PASS (0.0s)
    - title: Retrieval needs owned files
    - type: insight
    - excerpt: Local-first knowledge tools should retrieve from files the user already owns.
  4. audit: PASS (0.0s)

demo PASS — total 1.0s. This is your proof: agents search files you own.
Next: connect your own vault with exomem setup


Runs fully local and read-only against a sample vault bundled in the package.
Add `--keep` to leave that copy on disk afterward and open it in Obsidian.

## Install in one line

**Mac / Linux** — paste into Terminal:```bash
curl -fsSL https://raw.githubusercontent.com/Artexis10/exomem/main/scripts/install.sh | sh

Windows — paste into PowerShell:```powershell
irm https://raw.githubusercontent.com/Artexis10/exomem/main/scripts/install.ps1 | iex


Either one installs `uv`, installs exomem, and walks you through `exomem setup`,
which wires up **every agent client on the machine** — Claude Code and Codex both
get the MCP server, all ten skills, and the capture/retrieve hooks.

Safe to run again later — it skips whatever's already done. If it can't prompt
you interactively (e.g. run from another script), it prints the exact command to
run next instead of guessing. Prefer to run each step yourself? See below or the
full manual walkthrough in
QUICKSTART.md.

### Claude Code: install as a plugin

The plugin carries the skills and hooks. Its MCP entry is an optional shared
HTTP URL; it deliberately does **not** auto-start a full stdio server in every
Claude Code session:```
/plugin marketplace add Artexis10/exomem
/plugin install exomem@exomem

Then choose the route. If you already run Exomem as a service for phone or
hosted access, reuse that one authenticated core:```bash
exomem setup --vault "/path/to/vault" --mcp-url https://<host>/mcp


For local-only use, run `exomem setup --vault &quot;/path/to/vault&quot; --stdio`.
Stdio is intentionally explicit because each client session owns its process
and can load its own embedding runtime. When updating from an older plugin,
run `/plugin update exomem@exomem`, then `/reload-plugins` or restart Claude
Code before rerunning setup.

### claude.ai and ChatGPT

Neither platform can install skills programmatically — upload is manual. Build
the archives, then upload them in the client&#x27;s settings:```bash
exomem package-skills          # writes dist/skills/*.zip (all ten)

Point the client at your server as a connector (see
docs/remote-quickstart.md). Neither platform has a
hook system, so capture there is skill-driven rather than automatic.

Set it up in 5 minutes```bash

uv tool install exomem # or: pip install exomem
exomem setup --vault "/path/to/your/Obsidian"


Re-running the one-line installer upgrades an existing uv-tool install instead
of leaving an older `exomem` command on PATH. `exomem --version --json` is a
model-free provenance check, so it also works in the intentional lean CLI paired
with a fuller managed service.

One command does the whole local setup: the wizard scans your vault and shows
what&#x27;s already there, initializes `Knowledge Base/`, runs the `doctor`
preflight, registers the server with every client it detects (Claude Code and
Codex), and installs the skills into each.

If cwd `.env` or the process contains `EXOMEM_BASE_URL`, setup registers its
canonical `/mcp` endpoint with Claude Code and Codex so they share the existing
service. Otherwise it uses the explicit local stdio fallback. Use `--mcp-url`
to choose a service directly, `--stdio` to force local stdio, and
`--replace-client-registration` when you deliberately want to replace an
existing manual route. Native clients handle OAuth; setup never stores bearer
tokens.

Config files it did not create are treated as yours: it merges rather than
overwrites, backs the file up first, and prints the diff.

Already have a vault full of notes? That&#x27;s the normal case: `adopt` gives a
scan-first, read-only report of what&#x27;s there, suggested knowledge packs, and
safe copy/compile-planning next actions. Exomem only ever writes under `Knowledge Base/` — your
existing files stay untouched unless you explicitly copy or compile selected material. See
QUICKSTART.md § Already have a vault full of notes?
for the full contract, including daily-notes vaults. Re-running `setup` is
safe; completed steps report `[skipped]`. Non-interactive:
`exomem setup --yes --vault &quot;/path&quot; --lean`.

The individual steps (`exomem init` / `doctor` / `install-skill` /
`install-hook`, plus `claude mcp add`) still exist as the manual path — see
QUICKSTART.md.

The skill installs under the Claude Code name `exomem` — the same name as the
connector, so skill, server, and tools all read as one product. The skill is
recommended for Claude Code —
the server gives Claude the tools, the skill is what makes it use them. Hooks
are local-client reliability helpers for Claude Code and Codex: a read-side
reminder before answers, a write-side reminder at natural stopping points, and
a shared local continuation checkpoint around compaction. The checkpoint stores
bounded structural evidence (repository state, artifact hashes/counts, and
transcript provenance hashes), never conversation, tool, summary, or artifact
content. It needs no MCP connection or Exomem credential.
The read-side hook suppresses obvious control/status prompts like `continue`,
`merge it`, and `are you done?`, and can optionally upgrade that reminder to real
retrieved KB content (`EXOMEM_RETRIEVE_INJECT=1`, opt-in; the legacy
`KB_RETRIEVE_INJECT` name still works) — or, with
`EXOMEM_RETRIEVE_INJECT=working_set`, to a compiled `recall_context` packet
under a data header, bounded by `EXOMEM_RETRIEVE_INJECT_MAX_CHARS` (default
4,000). For Codex, run
`exomem install-hook --client codex`; for Claude Code, `exomem install-hook` —
see
QUICKSTART.md § 7.
Both clients checkpoint on manual or automatic `PreCompact` and reinject on
`SessionStart(compact|resume)`; Claude also checkpoints on `SessionEnd`. Pinned
Codex CLI 0.144.3 has no `SessionEnd` hook, so Exomem does not invent one.
Other MCP clients can still use the server. If they do not support Skills,
have them call `bootstrap()` once at the start of the session; it returns the
same compact operating contract through MCP, including when to search, when to
save, workflow-skill discovery, upload guidance, and performance profiles. It
also teaches the authoring loop: search first, draft the typed note, run
`suggest_links`, write with the right tool, inspect warnings/suggestions, then
report the path.

How eagerly it does that is tunable: four prominence levels (`off` / `light` /
`balanced` / `maximal`), set with `exomem prominence ` or pasted into a web
assistant&#x27;s custom instructions. Clients with hooks default to `balanced`; web and
hosted default to `maximal`, because nothing there re-arms the check between turns.
See docs/prominence.md for the levels and the copy-paste
blocks.

For client-specific assistant instructions, see
docs/ai-assistant-guide.md. For the boundary
between Exomem and a chat product&#x27;s built-in memory, see
docs/vs-built-in-memory.md.

Full local setup is in QUICKSTART.md. Remote/mobile setup is
in docs/remote-quickstart.md and
docs/deployment.md.

Remote OAuth uses GitHub once to prove the configured login plus immutable user
ID, then issues an Exomem-owned durable session. Operators can inspect and revoke
those sessions with `exomem auth sessions`, `exomem auth revoke `,
and `exomem auth revoke --all`; session administration is never an MCP tool.

For development, or to run the sample vault from a checkout instead of a
package install:```bash
git clone https://github.com/Artexis10/exomem.git
cd exomem
uv sync
uv run exomem demo

Connect your agent

Client How
Claude Code exomem setup registers it for you (see above)
Codex CLI codex mcp add plus optional exomem install-hook --client codex - see docs/ai-assistant-guide.md#codex-cli
claude.ai or hosted chat Remote MCP/connector - see docs/remote-quickstart.md and docs/ai-assistant-guide.md#hosted-chat-clients
Any MCP client Generic stdio config - see below and docs/ai-assistant-guide.md#generic-stdio-mcp-clients; call bootstrap() first
Docker (no Python) One docker run line — see below and docs/docker.md

Codex CLI```bash
codex mcp add exomem --env EXOMEM_VAULT_PATH="/path/to/vault" -- exomem --transport stdio


Optional local hooks, using the same Exomem scripts as Claude Code:```bash
exomem install-hook --client codex

Verify deployed Claude Code/Codex hooks without changing anything:```bash
exomem install-hook --check


For the `AGENTS.md` instruction block and the &quot;do not search every tiny prompt&quot;
policy, see docs/ai-assistant-guide.md.

Or add it directly to `~/.codex/config.toml`:```toml
[mcp_servers.exomem]
command = &quot;exomem&quot;
args = [&quot;--transport&quot;, &quot;stdio&quot;]
env = { EXOMEM_VAULT_PATH = &quot;/path/to/vault&quot; }

Any MCP client (generic stdio)```json
{"mcpServers": {"exomem": {"command": "exomem", "args": ["--transport", "stdio"], "env": {"EXOMEM_VAULT_PATH": "/path/to/vault"}}}}


After connecting, ask the agent to call `bootstrap()` before using the KB. Claude
Skills are still the best UX where available, but `bootstrap()` lets generic MCP
clients learn Exomem&#x27;s search/save/upload contract without a separate skill file,
including the write loop for compiled notes.
See docs/ai-assistant-guide.md for the copyable
standing instruction.

Docker (no Python on the host)```bash
claude mcp add exomem -- docker run -i --rm -v &quot;/path/to/vault:/vault&quot; -e EXOMEM_VAULT_PATH=/vault ghcr.io/artexis10/exomem:latest --transport stdio

Use :latest for the lean keyword/BM25 image, :ml for CPU hybrid search, or
:cuda for NVIDIA/Linux CUDA capability. CUDA images still boot CPU-default at
idle; opt into GPU residency with EXOMEM_MODE=performance when you want it.
The image also runs as an always-on remote server via docker compose with a
tunnel sidecar — see docs/docker.md. Windows users with a live
vault should usually prefer the native install (WSL2 bind mounts miss live
file-watch events). macOS cannot serve a vault today: exomem's held-filesystem
substrate, which every governed write acquires a reserved-path root through, has
a Linux backend and a Windows backend and no darwin implementation, so exomem
refuses there and exomem doctor says why.

The first start downloads search models in the background — find works
immediately with keyword ranking and upgrades to semantic search automatically
once the models land. Run exomem warm to pre-download them ahead of time.

Resource modes

Exomem is CPU-first by default so an idle server does not quietly occupy GPU
memory. Control the machine footprint without editing code:```bash
exomem mode quiet # low-resource: no heavy warm-up, evict caches, defer semantic reindex
exomem mode normal # default: CPU steady-state, warm CPU caches allowed
exomem mode performance # explicit opt-in for GPU-capable bulk/model work
exomem status --resources --json


Use `quiet` before gaming or other foreground workloads. Keyword/BM25 freshness,
file-change freshness, inbound links, and resolver state stay live; expensive
semantic/CLIP reindex work can be deferred and is reported in resource status.
Run `exomem index` or `kb reconcile` later to heal deferred semantic work.

Automatic quiet switching is optional. Its watcher skips GPU pressure queries
when you select `quiet` manually, pin a mode with `EXOMEM_MODE`, or disable
`EXOMEM_AUTO_QUIET`. Automatic quiet still monitors pressure so it can restore
the previous mode.

On affected WSL/NVIDIA systems, repeated Linux `nvidia-smi` processes can grow
the Windows driver&#x27;s kernel memory and disrupt foreground work.
Microsoft WSL issue #12756
records a reproduction. For older Exomem versions, a service configured with
`EXOMEM_AUTO_QUIET=0` stops its watcher after that service restarts.
An empty `CUDA_VISIBLE_DEVICES` also disables GPU work and GPU headroom queries.
These settings do not require a Windows reboot; existing driver state may remain.

## What it does

- **Searches the vault you already own.** Markdown stays in place; exomem does
  not import copies into a proprietary note store.
- **Adopts messy vaults safely.** `adopt` starts with a read-only report and
  explicit copy and compile-planning options, so originals remain archival until you choose.
- **Retrieves across text and media.** Markdown, PDFs, Office docs, images,
  screenshots, audio, and video can become searchable through local extraction.
- **Keeps sources separate from conclusions.** Raw captures, compiled notes,
  entities, evidence, and superseded conclusions live in typed folders.
- **Surfaces review work.** Audit and attention queues can show unprocessed
  sources, stale notes, broken links, and close-by claims worth reviewing.
- **Measures, never judges.** The server does deterministic work: search,
  extraction, ranking, embeddings, file writes, and graph checks. Reasoning stays
  in the client model.

## Why use it

Most AI note tools make you move into their app or ingest your files into their
store. exomem works the other way around: agents come to your vault.

| Compared with | Difference |
| --- | --- |
| Doc-chat / RAG apps | exomem works over live files instead of imported copies. |
| Basic MCP note servers | exomem adds typed knowledge operations, multimodal extraction, audit queues, and CLI/REST parity. |
| Memory hidden inside one assistant | exomem is client-agnostic: use the same vault from Claude Code, Codex, Cursor, scripts, or a custom chatbot. |

For a deeper point-in-time comparison, see
docs/comparison-engraph.md. The reproducible
graph-only comparison with Basic Memory is in
docs/comparison-basic-memory-graph.md.
For the practical boundary with chat products&#x27; own memory features, see
docs/vs-built-in-memory.md.

**Measured retrieval quality — and speed.** Retrieval is graded by a
reproducible golden-set eval harness, not asserted, and latency is measured
per lane at corpus scale: hybrid `find()` runs sub-second end-to-end at
50,000 notes (864 ms on the reference desktop, hot cache off), with the
keyword/BM25 lanes served from an FTS5 sidecar index in milliseconds —
built into stdlib SQLite, so it works on the lean install too. Methodology
and numbers in docs/benchmarks.md.

## Simple front door

Agents should route normal user requests through simple actions first, then use the typed tools underneath.

| Action | Use when the user says | Backed by |
| --- | --- | --- |
| Save | &quot;remember this&quot;, &quot;log this&quot;, &quot;this is a decision&quot; | `add`, `note`, `link`, `preserve` |
| Adopt/import | &quot;make this old vault usable&quot;, &quot;import my notes safely&quot; | `adopt`, `overview` |
| Ask | &quot;what do we know about X?&quot;, &quot;show the sources&quot; | `find`, `get` |
| Prove | &quot;save this for the warranty case&quot;, &quot;show the evidence&quot; | `preserve`, upload/download, `find` |
| Review | &quot;what needs cleanup?&quot;, &quot;what is stale?&quot; | `attention`, `audit`, `propose_compilation` |
| Update | &quot;this replaced the old conclusion&quot;, &quot;fix that note&quot; | `edit`, `replace`, `reconcile` |
| Connect | &quot;link this to X&quot;, &quot;what should this cite?&quot; | `link`, `suggest_links` |
| Plan | &quot;save this feature idea&quot;, &quot;file this bug for later&quot;, &quot;what matters this week&quot; | `plan_memory` with `inspect`, `create`, `query`, `add`, `update`, or `triage` |
| Record | &quot;log this session&quot;, &quot;record this measurement&quot;, &quot;update the mileage&quot; | `record_memory` with `inspect`, `create`, `query`, `append`, or `update` |

## Core tools

exomem exposes typed MCP tools for common knowledge-base work:

| Tool | Purpose |
| --- | --- |
| `recall_context` | Compile durable context for a raw turn, with no query: returns a map or resolves which anchors the turn is about, with a bounded working-memory packet, or abstains. |
| `episode_memory` | Record a bounded recap of what a conversation worked on, decided and left open, so the next session on any client sees it first; inspect an episode&#x27;s revisions. |
| `find` | Search notes, sources, entities, and evidence with type/project/tag filters. |
| `get` | Read a full page or frontmatter. |
| `add` | Capture a raw source page. |
| `note` | Create compiled notes: research note, insight, failure, pattern, experiment, or production log. |
| `edit` | Patch an existing compiled page. |
| `replace` | Supersede an old conclusion with a new one and preserve the link between them. |
| `preserve` | Store binary or text evidence append-only. |
| `audit` | Check graph and corpus health. |
| `attention` | Surface review queues such as stale notes, close-by claims, and unprocessed sources. |
| `overview` | Bounded, read-only structure report of the vault or a subtree — works outside `Knowledge Base/` and before `init`. |
| `adopt` | Existing-vault adoption: scan-only by default; can save a manifest, copy selected legacy text files as Sources, or return a compile plan while preserving originals. |
| `record_memory` | One front door for human-owned observed state: inspect, create, query, append, or targeted update a governed Record collection. |
| `plan_memory` | One front door for human-owned intended future state: inspect, create, query, add, update, or triage a Planning collection. |

Tier-2 tools cover governance administration and filesystem escape hatches such
as listing directories, creating files, moving pages, trashing files, and
recovering from trash. Set
`EXOMEM_DISABLE_TIER2=1` if you want a smaller tool surface.

Every write records durable history in `Knowledge Base/log.md`. Service calls
also go to `logs/exomem.log`.

## One operation, three doors

Every operation is declared once and exposed through:

- **MCP** for agents.
- **CLI** for terminal and scripts.
- **REST** for personal HTTP integrations when `EXOMEM_REST_API_KEY` is set.

Examples:```bash
kb find &quot;project handoff&quot; --mode keyword
kb find &quot;stale decision&quot; --json
kb get &quot;Notes/Insights/retrieval-needs-owned-files&quot; --json
kb note --note-type insight --title &quot;Agents need durable context&quot; \
  --content &quot;# Agents need durable context&quot;
``````bash
curl -s -X POST http://127.0.0.1:8765/api/find \
  -H &quot;Authorization: Bearer $EXOMEM_REST_API_KEY&quot; \
  -H &quot;Content-Type: application/json&quot; \
  -d &#x27;{&quot;query&quot;: &quot;project handoff&quot;, &quot;mode&quot;: &quot;keyword&quot;}&#x27;

CLI and REST share the same JSON envelope:```json
{"success": true, "data": []}


## Multimodal by default, resident only while working

The native release-service command installs the `standard` profile: local
embeddings, PDFs/Office documents, OCR bindings, ASR, and CLIP. Heavy models are
not loaded at service startup. Media jobs enter a durable queue, run serially in
one disposable child process, and the child exits after five idle minutes so its
RAM and MPS/MLX/CUDA state return to the host.

Manual/source installs can select the same capabilities directly:```bash
uv sync --extra embeddings
uv sync --extra media
  • embeddings: local text embeddings plus CLIP image search.
  • media: OCR for images, PDF extraction, Office document extraction, and
    faster-whisper ASR for audio/video.

CSV, TSV and JSON evidence uploads get dataset source cards for discovery and
exact queries. YAML, JSONL, XML and TOML exports get bounded text previews.
These paths need no optional extras; see source formats
for limits and governance behavior.

lean and hybrid remain available for constrained machines. Generated image
captioning and speaker diarization are advanced opt-ins, not requirements for
the standard multimodal path.

System tools: Tesseract is required for image OCR. On Windows:powershell winget install --id UB-Mannheim.TesseractOCR -e

GPU acceleration is useful but not required. Steady-state torch models default to
CPU in normal and quiet; performance is the explicit opt-in for capable
NVIDIA CUDA or Apple Silicon MPS/Metal paths. See
docs/deployment.md for CUDA, Blackwell, Apple Silicon,
diarization, and remote-service details.

Configuration

The server reads environment variables or a .env file. The main ones are:

Variable Purpose
EXOMEM_VAULT_PATH Vault root containing the governed folder (default Knowledge Base/).
EXOMEM_KB_DIRNAME Name of the governed folder inside the vault (default Knowledge Base).
EXOMEM_DISABLE_EMBEDDINGS 1 forces keyword/BM25-only search.
EXOMEM_RECALL_MODEL Names the recall encoder. Default: BAAI/bge-m3 on a personal server and a cloud cell, BAAI/bge-base-en-v1.5 on a hosted cell. A vault whose vectors came from another model is re-embedded in the background.
EXOMEM_RECALL_REEMBED off builds no new sidecar. A personal server keeps serving the old sidecar with its model; a cell keeps dense recall off.
EXOMEM_DISABLE_TIER2 1 hides Tier-2 filesystem tools.
EXOMEM_REST_API_KEY Enables authenticated REST routes.
EXOMEM_DISABLE_MEDIA_EXTRACTION 1 skips server-side OCR/ASR/PDF/Office extraction.
EXOMEM_DISABLE_CLIP 1 disables CLIP image search.
EXOMEM_MODE Hard-pin resource mode: quiet, normal, or performance. Env wins over config.
EXOMEM_QUIET_MODE Legacy truthy alias for quiet when EXOMEM_MODE is unset.
EXOMEM_AUTO_QUIET 1 enables optional non-torch GPU-pressure auto-quiet switching (default off).
EXOMEM_DEVICE / EXOMEM_TORCH_DEVICE Force all torch models to cuda, mps, or cpu. Normally leave unset and use exomem mode.
EXOMEM_MPS_FP16 On Apple Silicon, run bge/CLIP in fp16 on the Metal GPU — ~half the memory, faster encodes (default on; set 0 to keep fp32).
EXOMEM_VIDEO_SCENE_FRAMES Set to enable video scene detection + persisted, OCR'd scene-frame JPEGs (default off).
EXOMEM_VIDEO_SCENE_THRESHOLD Scene-boundary hash threshold in bits of 64 (default 10).
EXOMEM_VIDEO_SCENE_MIN_SECS Minimum scene duration in seconds; closer boundaries merge (default 4).
EXOMEM_SEMANTIC_SEGMENTS Set to enable timed transcripts + semantic segment retrieval for audio/video (default off).
EXOMEM_WHISPER_MODEL faster-whisper model size for ASR, such as base or small.
EXOMEM_ASR_BACKEND ASR engine: mlx (Apple Silicon Metal GPU, needs the media-mlx extra) or faster-whisper (CUDA/CPU). Default auto-selects MLX on Apple Silicon, else faster-whisper.
EXOMEM_MLX_WHISPER_MODEL HF repo for the MLX ASR model (default mlx-community/whisper-large-v3-mlx; use mlx-community/whisper-large-v3-turbo for speed).
EXOMEM_TESSERACT_CMD Path to the tesseract binary if not auto-discovered.
EXOMEM_CLAIM_LEVEL 1 enables the claim-level subsystem (default off).
EXOMEM_CLAIM_POLARITY_NLI 1 opts into the frozen stance verifier (default off). It still runs only if the exact repository-pinned revision and declared artifact manifest are resident and verified.
EXOMEM_CONTRADICTION_TOP_N Caps the surfaced contradiction queue and therefore verifier enrichment (default 40; 0 is uncapped). This is the sole polarity-work bound.

EXOMEM_CLAIM_NLI_MODEL is retired. Model identity comes only from the
in-repo pin registry; a value left in it selects nothing and is reported as
ignored by exomem doctor.

The shipped pin is the multilingual
MoritzLaurer/mDeBERTa-v3-base-xnli-multilingual-nli-2mil7
checkpoint at one exact upstream revision. Its admission fixtures check English,
German, French, Estonian, and mixed English/Estonian examples. That is bounded
evidence for those examples, not a promise of equal quality across every language
in the model card. Its labels are NLI relations: contradict, duplicate,
refine, or neutral. neutral includes unrelated, compatible, and uncertain
pairs; it never means “proved unrelated.”

Degradation modes

Optional tiers are absent by default and degrade to silence, not to a
substitute. What "absent" means, per tier:

Tier Install When absent
Semantic search --extra embeddings Keyword/BM25 retrieval; every surface still answers.
Media extraction --extra media Server-side OCR/ASR/PDF extraction is skipped; the model-driven upload text path still works.
Vector KNN (sqlite-vec) with embeddings Exact in-memory scan over the same vectors; identical results, more work.
Frozen stance verifier --extra nli Review-queue entries carry no model polarity label — never a differently-produced label under the verifier's name. Write responses, rankings, and every other surface are byte-identical to a build with no verifier tier at all; only exomem doctor names the absence. The verifier runs only under an exact pinned revision and artifact-manifest digest, a versioned label map, and a green bounded multilingual fixture set, and it labels review-queue entries only. Hosted cells do not install or enable it.

Legacy EXOMEM_* names (from the project's former working name, exomem) remain
honored: each is promoted to its EXOMEM_* equivalent at startup, with an
explicitly set EXOMEM_* value winning on conflict. import exomem and
python -m exomem likewise keep working as deprecated aliases.

Remote-only variables and full deployment notes are in
docs/deployment.md.

Project status

exomem is packaged on PyPI, uses Release Please for versioning, and follows the
lightweight SemVer policy in docs/release.md. The public CLI
entry point is exomem; kb is the short daily-driver alias for knowledge-base
operations.

License

AGPL-3.0-or-later. See LICENSE.

常见问题

6 问
答案由本站依据条目字段整理,不引用上游原文
Exomem 怎么安装?

资料没有给出安装命令,请到官方仓库或文档按说明接入;这类条目多数是生态应用,或尚未发布可安装包。

Exomem 怎么在 Claude Desktop、Claude、Cursor 里配置?

它的传输方式未标注,在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,配置才会生效。

Exomem 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

Exomem 需要 API Key 吗?

资料没有标注鉴权方式,是否需要密钥取决于服务本身,接入前按官方文档确认。

Exomem 支持哪些客户端?

资料标注支持 Claude Desktop、Claude、Cursor、VS Code、Cline、Codex;其他客户端只要能配置 MCP Server,也能用同样的命令或地址接入。

Exomem 能提供哪些能力?

资料显示能力标签包括 记忆。具体工具清单以官方文档为准。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 未标注。