orimem-mcp 0.1.1__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- orimem_mcp-0.1.1/LICENSE +21 -0
- orimem_mcp-0.1.1/PKG-INFO +263 -0
- orimem_mcp-0.1.1/README.md +224 -0
- orimem_mcp-0.1.1/pyproject.toml +57 -0
- orimem_mcp-0.1.1/setup.cfg +4 -0
- orimem_mcp-0.1.1/src/orimem/__init__.py +1 -0
- orimem_mcp-0.1.1/src/orimem/__main__.py +4 -0
- orimem_mcp-0.1.1/src/orimem/adapters/__init__.py +431 -0
- orimem_mcp-0.1.1/src/orimem/capture/__init__.py +110 -0
- orimem_mcp-0.1.1/src/orimem/capture/claude.py +150 -0
- orimem_mcp-0.1.1/src/orimem/capture/codex.py +116 -0
- orimem_mcp-0.1.1/src/orimem/capture/common.py +145 -0
- orimem_mcp-0.1.1/src/orimem/capture/dsh.py +153 -0
- orimem_mcp-0.1.1/src/orimem/capture/worker.py +24 -0
- orimem_mcp-0.1.1/src/orimem/cli.py +342 -0
- orimem_mcp-0.1.1/src/orimem/config.py +197 -0
- orimem_mcp-0.1.1/src/orimem/core/__init__.py +3 -0
- orimem_mcp-0.1.1/src/orimem/core/assessment_revision.py +147 -0
- orimem_mcp-0.1.1/src/orimem/core/backend.py +925 -0
- orimem_mcp-0.1.1/src/orimem/core/candidate_reviews.py +154 -0
- orimem_mcp-0.1.1/src/orimem/core/content_policy.py +19 -0
- orimem_mcp-0.1.1/src/orimem/core/continuous.py +202 -0
- orimem_mcp-0.1.1/src/orimem/core/corrections.py +265 -0
- orimem_mcp-0.1.1/src/orimem/core/embedding.py +189 -0
- orimem_mcp-0.1.1/src/orimem/core/evidence_context.py +143 -0
- orimem_mcp-0.1.1/src/orimem/core/evidence_pack.py +56 -0
- orimem_mcp-0.1.1/src/orimem/core/history.py +288 -0
- orimem_mcp-0.1.1/src/orimem/core/knowledge_index.py +187 -0
- orimem_mcp-0.1.1/src/orimem/core/lifecycle.py +729 -0
- orimem_mcp-0.1.1/src/orimem/core/markdown.py +123 -0
- orimem_mcp-0.1.1/src/orimem/core/mcp_server.py +362 -0
- orimem_mcp-0.1.1/src/orimem/core/message_sources.py +54 -0
- orimem_mcp-0.1.1/src/orimem/core/migration.py +466 -0
- orimem_mcp-0.1.1/src/orimem/core/pending_index.py +212 -0
- orimem_mcp-0.1.1/src/orimem/core/privacy.py +129 -0
- orimem_mcp-0.1.1/src/orimem/core/records.py +144 -0
- orimem_mcp-0.1.1/src/orimem/core/retrieval_policy.py +177 -0
- orimem_mcp-0.1.1/src/orimem/core/service.py +178 -0
- orimem_mcp-0.1.1/src/orimem/core/source_index.py +113 -0
- orimem_mcp-0.1.1/src/orimem/core/source_stage.py +144 -0
- orimem_mcp-0.1.1/src/orimem/core/stage_jobs.py +155 -0
- orimem_mcp-0.1.1/src/orimem/core/task_context.py +501 -0
- orimem_mcp-0.1.1/src/orimem/core/watcher.py +98 -0
- orimem_mcp-0.1.1/src/orimem/daily.py +113 -0
- orimem_mcp-0.1.1/src/orimem/hook.py +122 -0
- orimem_mcp-0.1.1/src/orimem/hook_entry.py +106 -0
- orimem_mcp-0.1.1/src/orimem/journal_state.py +487 -0
- orimem_mcp-0.1.1/src/orimem/journal_writer.py +465 -0
- orimem_mcp-0.1.1/src/orimem/lifecycle_ctl.py +24 -0
- orimem_mcp-0.1.1/src/orimem/model.py +140 -0
- orimem_mcp-0.1.1/src/orimem/skill/orimem/SKILL.md +34 -0
- orimem_mcp-0.1.1/src/orimem/skill/orimem/agents/openai.yaml +3 -0
- orimem_mcp-0.1.1/src/orimem/skill/orimem/references/automatic-journaling.md +111 -0
- orimem_mcp-0.1.1/src/orimem/skill/orimem/references/history.md +37 -0
- orimem_mcp-0.1.1/src/orimem/skill/orimem/references/lifecycle.md +91 -0
- orimem_mcp-0.1.1/src/orimem/skill/orimem/references/maintenance.md +18 -0
- orimem_mcp-0.1.1/src/orimem/skill/orimem/references/organizing.md +86 -0
- orimem_mcp-0.1.1/src/orimem/skill/orimem/references/retrieval.md +61 -0
- orimem_mcp-0.1.1/src/orimem/templates/vault/AGENTS.md +24 -0
- orimem_mcp-0.1.1/src/orimem_mcp.egg-info/PKG-INFO +263 -0
- orimem_mcp-0.1.1/src/orimem_mcp.egg-info/SOURCES.txt +66 -0
- orimem_mcp-0.1.1/src/orimem_mcp.egg-info/dependency_links.txt +1 -0
- orimem_mcp-0.1.1/src/orimem_mcp.egg-info/entry_points.txt +2 -0
- orimem_mcp-0.1.1/src/orimem_mcp.egg-info/requires.txt +15 -0
- orimem_mcp-0.1.1/src/orimem_mcp.egg-info/top_level.txt +1 -0
- orimem_mcp-0.1.1/tests/test_capture.py +87 -0
- orimem_mcp-0.1.1/tests/test_config_adapters.py +180 -0
- orimem_mcp-0.1.1/tests/test_release.py +35 -0
orimem_mcp-0.1.1/LICENSE
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
1
|
+
MIT License
|
|
2
|
+
|
|
3
|
+
Copyright (c) 2026 Orion
|
|
4
|
+
|
|
5
|
+
Permission is hereby granted, free of charge, to any person obtaining a copy
|
|
6
|
+
of this software and associated documentation files (the "Software"), to deal
|
|
7
|
+
in the Software without restriction, including without limitation the rights
|
|
8
|
+
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
|
|
9
|
+
copies of the Software, and to permit persons to whom the Software is
|
|
10
|
+
furnished to do so, subject to the following conditions:
|
|
11
|
+
|
|
12
|
+
The above copyright notice and this permission notice shall be included in all
|
|
13
|
+
copies or substantial portions of the Software.
|
|
14
|
+
|
|
15
|
+
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
|
|
16
|
+
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
|
|
17
|
+
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
|
|
18
|
+
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
|
|
19
|
+
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
|
|
20
|
+
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
|
|
21
|
+
SOFTWARE.
|
|
@@ -0,0 +1,263 @@
|
|
|
1
|
+
Metadata-Version: 2.4
|
|
2
|
+
Name: orimem-mcp
|
|
3
|
+
Version: 0.1.1
|
|
4
|
+
Summary: Local Obsidian-vault memory for coding agents (Claude Code, Codex, DeepSeek Harness) over MCP: hybrid recall, verified write-back, conversation capture and daily summaries.
|
|
5
|
+
Author: Orion
|
|
6
|
+
License: MIT
|
|
7
|
+
Project-URL: Homepage, https://github.com/Wen5555/OriMem
|
|
8
|
+
Project-URL: Documentation, https://github.com/Wen5555/OriMem#readme
|
|
9
|
+
Project-URL: Issues, https://github.com/Wen5555/OriMem/issues
|
|
10
|
+
Keywords: mcp,obsidian,memory,claude-code,codex,deepseek,agent,rag
|
|
11
|
+
Classifier: Development Status :: 3 - Alpha
|
|
12
|
+
Classifier: Environment :: Console
|
|
13
|
+
Classifier: Intended Audience :: Developers
|
|
14
|
+
Classifier: License :: OSI Approved :: MIT License
|
|
15
|
+
Classifier: Operating System :: Microsoft :: Windows
|
|
16
|
+
Classifier: Operating System :: MacOS
|
|
17
|
+
Classifier: Operating System :: POSIX :: Linux
|
|
18
|
+
Classifier: Programming Language :: Python :: 3.11
|
|
19
|
+
Classifier: Programming Language :: Python :: 3.12
|
|
20
|
+
Classifier: Programming Language :: Python :: 3.13
|
|
21
|
+
Classifier: Topic :: Software Development
|
|
22
|
+
Requires-Python: >=3.11
|
|
23
|
+
Description-Content-Type: text/markdown
|
|
24
|
+
License-File: LICENSE
|
|
25
|
+
Requires-Dist: numpy>=1.24
|
|
26
|
+
Requires-Dist: jsonschema>=4
|
|
27
|
+
Requires-Dist: pyyaml>=6
|
|
28
|
+
Requires-Dist: watchdog>=3
|
|
29
|
+
Requires-Dist: requests>=2.28
|
|
30
|
+
Requires-Dist: zstandard>=0.21
|
|
31
|
+
Requires-Dist: onnxruntime>=1.17
|
|
32
|
+
Requires-Dist: tokenizers>=0.19
|
|
33
|
+
Provides-Extra: torch
|
|
34
|
+
Requires-Dist: torch>=2.1; extra == "torch"
|
|
35
|
+
Requires-Dist: transformers>=4.40; extra == "torch"
|
|
36
|
+
Provides-Extra: test
|
|
37
|
+
Requires-Dist: pytest>=7; extra == "test"
|
|
38
|
+
Dynamic: license-file
|
|
39
|
+
|
|
40
|
+
<div align="center">
|
|
41
|
+
|
|
42
|
+
<img src="assets/icon.png" width="96" alt="OriMem icon">
|
|
43
|
+
|
|
44
|
+
# OriMem
|
|
45
|
+
|
|
46
|
+
**Long-term memory for coding agents, kept in your own Obsidian vault.**
|
|
47
|
+
|
|
48
|
+
One local MCP server and one skill, shared by Claude Code, Codex and DeepSeek Harness —
|
|
49
|
+
recall project context across sessions, write confirmed results back with receipts, and capture every conversation automatically.
|
|
50
|
+
|
|
51
|
+
[](https://github.com/Wen5555/OriMem/actions/workflows/ci.yml)
|
|
52
|
+

|
|
53
|
+

|
|
54
|
+

|
|
55
|
+
[](LICENSE)
|
|
56
|
+
|
|
57
|
+
**English** · [简体中文](README.zh-CN.md)
|
|
58
|
+
|
|
59
|
+
</div>
|
|
60
|
+
|
|
61
|
+
<!-- mcp-name: io.github.Wen5555/orimem -->
|
|
62
|
+
|
|
63
|
+
---
|
|
64
|
+
|
|
65
|
+
## Why OriMem
|
|
66
|
+
|
|
67
|
+
Coding agents forget everything when a session ends, and each agent keeps its history in its own private format. OriMem gives them **one shared, human-readable memory**:
|
|
68
|
+
|
|
69
|
+
- 🗂️ **Your vault is the memory.** Plain Markdown in an Obsidian vault you can read, edit and back up. The index is just a cache.
|
|
70
|
+
- 🔎 **Hybrid recall.** Multilingual E5 embeddings + SQLite FTS5 + wiki-links. Short snippets first, exact evidence on demand, so context stays small.
|
|
71
|
+
- ✅ **Verified write-back.** `prepare_record → diff → commit_record` with hash checks, backups, read-back and receipts. It never silently overwrites a note you edited.
|
|
72
|
+
- 🎙️ **Automatic capture.** Hooks save each Claude Code / Codex / dsh session as a transcript note when a turn ends. Tool output, reasoning, injected system context, sub-agents and automations are left out.
|
|
73
|
+
- 📅 **Daily summary (opt-in).** The agent asks you once; if you agree, a scheduled job writes the day's journal through the same verified writer.
|
|
74
|
+
- 🔒 **Local only.** No cloud calls, no telemetry. The embedding model is pinned and checksum-verified, and runs on CPU through ONNX Runtime (no PyTorch needed).
|
|
75
|
+
|
|
76
|
+
## Quick start
|
|
77
|
+
|
|
78
|
+
**Windows (PowerShell)**
|
|
79
|
+
|
|
80
|
+
```powershell
|
|
81
|
+
irm https://raw.githubusercontent.com/Wen5555/OriMem/main/scripts/install.ps1 | iex
|
|
82
|
+
```
|
|
83
|
+
|
|
84
|
+
**macOS / Linux**
|
|
85
|
+
|
|
86
|
+
```bash
|
|
87
|
+
curl -LsSf https://raw.githubusercontent.com/Wen5555/OriMem/main/scripts/install.sh | sh
|
|
88
|
+
```
|
|
89
|
+
|
|
90
|
+
The installer sets up [uv](https://github.com/astral-sh/uv) if needed, installs OriMem into an isolated environment (about 120 MB, Python provided by uv), then runs `orimem init`. `init` asks three questions — vault path, conversation capture (default on), daily summary (default "ask me later") — downloads the 470 MB embedding model, and connects every agent it finds. Restart your agents afterwards.
|
|
91
|
+
|
|
92
|
+
Then just talk to your agent:
|
|
93
|
+
|
|
94
|
+
> *"What did we decide about the export feature last week?"*
|
|
95
|
+
> *"Record that the WebGPU renderer is done and the next step is drag-to-reorder layers."*
|
|
96
|
+
|
|
97
|
+
<details>
|
|
98
|
+
<summary>Other ways to install</summary>
|
|
99
|
+
|
|
100
|
+
```bash
|
|
101
|
+
uv tool install orimem-mcp
|
|
102
|
+
```
|
|
103
|
+
|
|
104
|
+
```bash
|
|
105
|
+
pipx install orimem-mcp
|
|
106
|
+
```
|
|
107
|
+
|
|
108
|
+
The PyPI distribution is `orimem-mcp`; the command it installs is `orimem`. To run the MCP server without installing it first, use `uvx --from orimem-mcp orimem mcp`.
|
|
109
|
+
|
|
110
|
+
```bash
|
|
111
|
+
orimem init
|
|
112
|
+
```
|
|
113
|
+
|
|
114
|
+
Non-interactive: `orimem init --yes --vault ~/Notes --agents claude,codex`. Behind a slow network, the installer switches to mirrors automatically when PyPI is unreachable (`ORIMEM_MIRROR=cn` forces it).
|
|
115
|
+
</details>
|
|
116
|
+
|
|
117
|
+
## Supported agents
|
|
118
|
+
|
|
119
|
+
| Agent | MCP server | Skill | Hooks (capture on turn end) | Transcripts read from |
|
|
120
|
+
|---|---|---|---|---|
|
|
121
|
+
| **Claude Code** | `claude mcp add -s user orimem` | `~/.claude/skills/orimem` | `~/.claude/settings.json` | `~/.claude/projects/*/<uuid>.jsonl` |
|
|
122
|
+
| **Codex** | `[mcp_servers.orimem]` in `~/.codex/config.toml` | `~/.codex/skills/orimem` | `~/.codex/hooks.json` | `~/.codex/sessions/**/rollout-*.jsonl` |
|
|
123
|
+
| **DeepSeek Harness (dsh)** | `dsh-mcp-client` entry in the profile's `cordis.patch.yml` | `~/.dsh/skills/orimem` | `dsh-hooks-claude-code` bridge | `~/.dsh/sessions/**/session.v4.jsonl.zstd` |
|
|
124
|
+
| Any other MCP client | `orimem snippet` prints the JSON | copy the skill folder | — | — |
|
|
125
|
+
|
|
126
|
+
**Claude Code users** can also install it as a plugin (MCP server, skill and hooks in one step; still needs the `orimem` command):
|
|
127
|
+
|
|
128
|
+
```text
|
|
129
|
+
/plugin install orimem --marketplace Wen5555/OriMem
|
|
130
|
+
```
|
|
131
|
+
|
|
132
|
+
On older Claude Code versions: `/plugin marketplace add Wen5555/OriMem`, then `/plugin install orimem@orimem`.
|
|
133
|
+
|
|
134
|
+
**DeepSeek Harness users** can also add the [`dsh-orimem`](integrations/dsh) bundle from the community plugin market (or `dsh plugin --profile <name> add dsh-orimem-<version>.tgz` from a [release](https://github.com/Wen5555/OriMem/releases)); it needs the `orimem` command from the quick start.
|
|
135
|
+
|
|
136
|
+
`orimem install` and `orimem uninstall` only add or remove OriMem's own entries, and every edited file is backed up first. Marker blocks in TOML/YAML files are removed byte for byte; JSON files keep all other content.
|
|
137
|
+
|
|
138
|
+
## How it works
|
|
139
|
+
|
|
140
|
+
```mermaid
|
|
141
|
+
flowchart LR
|
|
142
|
+
subgraph Agents
|
|
143
|
+
CC[Claude Code]
|
|
144
|
+
CX[Codex]
|
|
145
|
+
DS[DeepSeek Harness]
|
|
146
|
+
end
|
|
147
|
+
CC & CX & DS -- "MCP (stdio)" --> M[orimem mcp]
|
|
148
|
+
CC & CX & DS -. "Stop hook" .-> H[orimem hook]
|
|
149
|
+
M --> S[(local service<br/>loopback only)]
|
|
150
|
+
H -- "capture this session" --> S
|
|
151
|
+
S --> I[(SQLite: chunks · E5 vectors<br/>FTS5 · links · ledger)]
|
|
152
|
+
S <--> V[/"Obsidian vault<br/>(Markdown = source of truth)"/]
|
|
153
|
+
S -- "every 2 min" --> T[agent transcripts]
|
|
154
|
+
```
|
|
155
|
+
|
|
156
|
+
- **Recall:** `task_context` binds the current project; `recall` returns ranked snippets; `evidence` fetches exact source text with offsets and hashes.
|
|
157
|
+
- **Write-back:** the agent proposes a complete note change; OriMem checks the expected hash, backs up, writes, re-reads, verifies the claimed text and issues a receipt.
|
|
158
|
+
- **Capture:** per-agent parsers mirror ordinary user/assistant turns into one pipeline, which stages them and projects stable transcript notes under `90 归档/Codex佐证/`. Hooks hand the work to the running service (falling back to a detached capture process), so they also work inside sandboxed hook runners such as dsh's.
|
|
159
|
+
|
|
160
|
+
## Configuration
|
|
161
|
+
|
|
162
|
+
Settings live in `~/.orimem/config.toml`. Change them with `orimem config set <key> <value>` or ask your agent, which uses the `config_set` tool.
|
|
163
|
+
|
|
164
|
+
| Key | Default | Meaning |
|
|
165
|
+
|---|---|---|
|
|
166
|
+
| `vault` | `~/.orimem/vault` | Obsidian vault path |
|
|
167
|
+
| `capture.enabled` | `true` | Capture conversations into the vault |
|
|
168
|
+
| `capture.sources` | `["claude","codex","dsh"]` | Which agents to capture |
|
|
169
|
+
| `capture.since` | install time | Older sessions are not back-filled (`orimem capture --backfill` does it on request) |
|
|
170
|
+
| `capture.exclude_cwd` | `[]` | Never capture sessions started in these folders |
|
|
171
|
+
| `daily_summary.enabled` | `"ask"` | `"ask"` = ask once, off until answered · `true` · `false` |
|
|
172
|
+
| `daily_summary.time` | `"22:30"` | When the daily job runs |
|
|
173
|
+
| `daily_summary.agent` | auto | `claude` or `codex` runs the summary |
|
|
174
|
+
| `recall.hf_endpoint` | auto | Model download endpoint (falls back to hf-mirror.com) |
|
|
175
|
+
|
|
176
|
+
## Reference
|
|
177
|
+
|
|
178
|
+
<details>
|
|
179
|
+
<summary><b>CLI</b></summary>
|
|
180
|
+
|
|
181
|
+
| Command | What it does |
|
|
182
|
+
|---|---|
|
|
183
|
+
| `orimem init [-y]` | First-time setup |
|
|
184
|
+
| `orimem install / uninstall [--agent claude,codex,dsh] [--dsh-profile NAME]` | Connect or disconnect agents |
|
|
185
|
+
| `orimem doctor` | Check config, model, embedder, dependencies and agents |
|
|
186
|
+
| `orimem capture [--source X] [--session ID] [--backfill]` | Capture now |
|
|
187
|
+
| `orimem sync` · `orimem recall "query" --scope "10 项目"` | Index / search from the shell |
|
|
188
|
+
| `orimem daily status / run [--dry-run] / apply` | Daily summary |
|
|
189
|
+
| `orimem model status / download [--torch] / import DIR` | Embedding model |
|
|
190
|
+
| `orimem config get / set / path` | Settings |
|
|
191
|
+
| `orimem snippet` | MCP JSON + AGENTS.md snippet for other clients |
|
|
192
|
+
| `orimem --home DIR …` | Use another data directory (written into generated agent commands) |
|
|
193
|
+
</details>
|
|
194
|
+
|
|
195
|
+
<details>
|
|
196
|
+
<summary><b>MCP tools</b></summary>
|
|
197
|
+
|
|
198
|
+
| Tool | Purpose |
|
|
199
|
+
|---|---|
|
|
200
|
+
| `task_context` | Bind the current folder/task to a project entry |
|
|
201
|
+
| `recall` | Hybrid search (snippets) |
|
|
202
|
+
| `evidence` | Exact source text for result ids |
|
|
203
|
+
| `list_sessions` · `get_session_messages` | Browse captured conversations |
|
|
204
|
+
| `prepare_record` · `commit_record` | Verified write-back |
|
|
205
|
+
| `initialize` | Create a project/course entry |
|
|
206
|
+
| `lifecycle` | Source reading, candidates, corrections |
|
|
207
|
+
| `health` · `sync` · `archive_history` | Maintenance |
|
|
208
|
+
| `config_get` · `config_set` · `capture_now` | Settings and capture |
|
|
209
|
+
</details>
|
|
210
|
+
|
|
211
|
+
<details>
|
|
212
|
+
<summary><b>Vault layout</b></summary>
|
|
213
|
+
|
|
214
|
+
```text
|
|
215
|
+
00 收件箱/ inbox and knowledge candidates
|
|
216
|
+
10 项目/ one page per project (status, next action)
|
|
217
|
+
15 课程/ courses
|
|
218
|
+
30 知识/ reusable knowledge
|
|
219
|
+
50 日记/ 每日/ (daily) · 每周/ (weekly)
|
|
220
|
+
90 归档/ archive; 90 归档/Codex佐证/ holds captured transcripts
|
|
221
|
+
95 AI 工作区/ agent work notes
|
|
222
|
+
```
|
|
223
|
+
|
|
224
|
+
Writes are limited to `layout.folders`. `.obsidian` and files outside the vault are never read or modified.
|
|
225
|
+
</details>
|
|
226
|
+
|
|
227
|
+
## FAQ
|
|
228
|
+
|
|
229
|
+
**Do I need a GPU or PyTorch?** No. Embeddings run on CPU through ONNX Runtime. `orimem[torch]` plus `orimem model download --torch` enables the optional PyTorch backend, which produces identical vectors.
|
|
230
|
+
|
|
231
|
+
**Can I use my existing vault?** Yes. Point `vault` at it. OriMem only adds the folders it needs and never rewrites existing notes outside the verified write-back flow.
|
|
232
|
+
|
|
233
|
+
**I'm in mainland China.** Both installers detect an unreachable PyPI and switch to the Tsinghua PyPI mirror, the npmmirror Python builds and hf-mirror.com.
|
|
234
|
+
|
|
235
|
+
**How do I turn capture off for a project?** `orimem config set capture.exclude_cwd '["D:/work/secret"]'`. To turn it off everywhere: `orimem config set capture.enabled false`.
|
|
236
|
+
|
|
237
|
+
**How do I remove everything?** `orimem uninstall`, then `uv tool uninstall orimem-mcp`. Your vault and `~/.orimem` stay until you delete them.
|
|
238
|
+
|
|
239
|
+
## Roadmap
|
|
240
|
+
|
|
241
|
+
- [ ] Configurable / English vault layout
|
|
242
|
+
- [ ] More agents (Cursor, Gemini CLI, OpenCode)
|
|
243
|
+
- [ ] Optional quantized ONNX model for low-memory machines
|
|
244
|
+
|
|
245
|
+
## Contributing
|
|
246
|
+
|
|
247
|
+
```bash
|
|
248
|
+
git clone https://github.com/Wen5555/OriMem && cd OriMem
|
|
249
|
+
```
|
|
250
|
+
|
|
251
|
+
```bash
|
|
252
|
+
uv venv && uv pip install -e ".[test]"
|
|
253
|
+
```
|
|
254
|
+
|
|
255
|
+
```bash
|
|
256
|
+
uv run pytest
|
|
257
|
+
```
|
|
258
|
+
|
|
259
|
+
A manual end-to-end checklist for each agent is in [docs/TESTING.md](docs/TESTING.md). Issues and pull requests are welcome.
|
|
260
|
+
|
|
261
|
+
## License
|
|
262
|
+
|
|
263
|
+
[MIT](LICENSE)
|
|
@@ -0,0 +1,224 @@
|
|
|
1
|
+
<div align="center">
|
|
2
|
+
|
|
3
|
+
<img src="assets/icon.png" width="96" alt="OriMem icon">
|
|
4
|
+
|
|
5
|
+
# OriMem
|
|
6
|
+
|
|
7
|
+
**Long-term memory for coding agents, kept in your own Obsidian vault.**
|
|
8
|
+
|
|
9
|
+
One local MCP server and one skill, shared by Claude Code, Codex and DeepSeek Harness —
|
|
10
|
+
recall project context across sessions, write confirmed results back with receipts, and capture every conversation automatically.
|
|
11
|
+
|
|
12
|
+
[](https://github.com/Wen5555/OriMem/actions/workflows/ci.yml)
|
|
13
|
+

|
|
14
|
+

|
|
15
|
+

|
|
16
|
+
[](LICENSE)
|
|
17
|
+
|
|
18
|
+
**English** · [简体中文](README.zh-CN.md)
|
|
19
|
+
|
|
20
|
+
</div>
|
|
21
|
+
|
|
22
|
+
<!-- mcp-name: io.github.Wen5555/orimem -->
|
|
23
|
+
|
|
24
|
+
---
|
|
25
|
+
|
|
26
|
+
## Why OriMem
|
|
27
|
+
|
|
28
|
+
Coding agents forget everything when a session ends, and each agent keeps its history in its own private format. OriMem gives them **one shared, human-readable memory**:
|
|
29
|
+
|
|
30
|
+
- 🗂️ **Your vault is the memory.** Plain Markdown in an Obsidian vault you can read, edit and back up. The index is just a cache.
|
|
31
|
+
- 🔎 **Hybrid recall.** Multilingual E5 embeddings + SQLite FTS5 + wiki-links. Short snippets first, exact evidence on demand, so context stays small.
|
|
32
|
+
- ✅ **Verified write-back.** `prepare_record → diff → commit_record` with hash checks, backups, read-back and receipts. It never silently overwrites a note you edited.
|
|
33
|
+
- 🎙️ **Automatic capture.** Hooks save each Claude Code / Codex / dsh session as a transcript note when a turn ends. Tool output, reasoning, injected system context, sub-agents and automations are left out.
|
|
34
|
+
- 📅 **Daily summary (opt-in).** The agent asks you once; if you agree, a scheduled job writes the day's journal through the same verified writer.
|
|
35
|
+
- 🔒 **Local only.** No cloud calls, no telemetry. The embedding model is pinned and checksum-verified, and runs on CPU through ONNX Runtime (no PyTorch needed).
|
|
36
|
+
|
|
37
|
+
## Quick start
|
|
38
|
+
|
|
39
|
+
**Windows (PowerShell)**
|
|
40
|
+
|
|
41
|
+
```powershell
|
|
42
|
+
irm https://raw.githubusercontent.com/Wen5555/OriMem/main/scripts/install.ps1 | iex
|
|
43
|
+
```
|
|
44
|
+
|
|
45
|
+
**macOS / Linux**
|
|
46
|
+
|
|
47
|
+
```bash
|
|
48
|
+
curl -LsSf https://raw.githubusercontent.com/Wen5555/OriMem/main/scripts/install.sh | sh
|
|
49
|
+
```
|
|
50
|
+
|
|
51
|
+
The installer sets up [uv](https://github.com/astral-sh/uv) if needed, installs OriMem into an isolated environment (about 120 MB, Python provided by uv), then runs `orimem init`. `init` asks three questions — vault path, conversation capture (default on), daily summary (default "ask me later") — downloads the 470 MB embedding model, and connects every agent it finds. Restart your agents afterwards.
|
|
52
|
+
|
|
53
|
+
Then just talk to your agent:
|
|
54
|
+
|
|
55
|
+
> *"What did we decide about the export feature last week?"*
|
|
56
|
+
> *"Record that the WebGPU renderer is done and the next step is drag-to-reorder layers."*
|
|
57
|
+
|
|
58
|
+
<details>
|
|
59
|
+
<summary>Other ways to install</summary>
|
|
60
|
+
|
|
61
|
+
```bash
|
|
62
|
+
uv tool install orimem-mcp
|
|
63
|
+
```
|
|
64
|
+
|
|
65
|
+
```bash
|
|
66
|
+
pipx install orimem-mcp
|
|
67
|
+
```
|
|
68
|
+
|
|
69
|
+
The PyPI distribution is `orimem-mcp`; the command it installs is `orimem`. To run the MCP server without installing it first, use `uvx --from orimem-mcp orimem mcp`.
|
|
70
|
+
|
|
71
|
+
```bash
|
|
72
|
+
orimem init
|
|
73
|
+
```
|
|
74
|
+
|
|
75
|
+
Non-interactive: `orimem init --yes --vault ~/Notes --agents claude,codex`. Behind a slow network, the installer switches to mirrors automatically when PyPI is unreachable (`ORIMEM_MIRROR=cn` forces it).
|
|
76
|
+
</details>
|
|
77
|
+
|
|
78
|
+
## Supported agents
|
|
79
|
+
|
|
80
|
+
| Agent | MCP server | Skill | Hooks (capture on turn end) | Transcripts read from |
|
|
81
|
+
|---|---|---|---|---|
|
|
82
|
+
| **Claude Code** | `claude mcp add -s user orimem` | `~/.claude/skills/orimem` | `~/.claude/settings.json` | `~/.claude/projects/*/<uuid>.jsonl` |
|
|
83
|
+
| **Codex** | `[mcp_servers.orimem]` in `~/.codex/config.toml` | `~/.codex/skills/orimem` | `~/.codex/hooks.json` | `~/.codex/sessions/**/rollout-*.jsonl` |
|
|
84
|
+
| **DeepSeek Harness (dsh)** | `dsh-mcp-client` entry in the profile's `cordis.patch.yml` | `~/.dsh/skills/orimem` | `dsh-hooks-claude-code` bridge | `~/.dsh/sessions/**/session.v4.jsonl.zstd` |
|
|
85
|
+
| Any other MCP client | `orimem snippet` prints the JSON | copy the skill folder | — | — |
|
|
86
|
+
|
|
87
|
+
**Claude Code users** can also install it as a plugin (MCP server, skill and hooks in one step; still needs the `orimem` command):
|
|
88
|
+
|
|
89
|
+
```text
|
|
90
|
+
/plugin install orimem --marketplace Wen5555/OriMem
|
|
91
|
+
```
|
|
92
|
+
|
|
93
|
+
On older Claude Code versions: `/plugin marketplace add Wen5555/OriMem`, then `/plugin install orimem@orimem`.
|
|
94
|
+
|
|
95
|
+
**DeepSeek Harness users** can also add the [`dsh-orimem`](integrations/dsh) bundle from the community plugin market (or `dsh plugin --profile <name> add dsh-orimem-<version>.tgz` from a [release](https://github.com/Wen5555/OriMem/releases)); it needs the `orimem` command from the quick start.
|
|
96
|
+
|
|
97
|
+
`orimem install` and `orimem uninstall` only add or remove OriMem's own entries, and every edited file is backed up first. Marker blocks in TOML/YAML files are removed byte for byte; JSON files keep all other content.
|
|
98
|
+
|
|
99
|
+
## How it works
|
|
100
|
+
|
|
101
|
+
```mermaid
|
|
102
|
+
flowchart LR
|
|
103
|
+
subgraph Agents
|
|
104
|
+
CC[Claude Code]
|
|
105
|
+
CX[Codex]
|
|
106
|
+
DS[DeepSeek Harness]
|
|
107
|
+
end
|
|
108
|
+
CC & CX & DS -- "MCP (stdio)" --> M[orimem mcp]
|
|
109
|
+
CC & CX & DS -. "Stop hook" .-> H[orimem hook]
|
|
110
|
+
M --> S[(local service<br/>loopback only)]
|
|
111
|
+
H -- "capture this session" --> S
|
|
112
|
+
S --> I[(SQLite: chunks · E5 vectors<br/>FTS5 · links · ledger)]
|
|
113
|
+
S <--> V[/"Obsidian vault<br/>(Markdown = source of truth)"/]
|
|
114
|
+
S -- "every 2 min" --> T[agent transcripts]
|
|
115
|
+
```
|
|
116
|
+
|
|
117
|
+
- **Recall:** `task_context` binds the current project; `recall` returns ranked snippets; `evidence` fetches exact source text with offsets and hashes.
|
|
118
|
+
- **Write-back:** the agent proposes a complete note change; OriMem checks the expected hash, backs up, writes, re-reads, verifies the claimed text and issues a receipt.
|
|
119
|
+
- **Capture:** per-agent parsers mirror ordinary user/assistant turns into one pipeline, which stages them and projects stable transcript notes under `90 归档/Codex佐证/`. Hooks hand the work to the running service (falling back to a detached capture process), so they also work inside sandboxed hook runners such as dsh's.
|
|
120
|
+
|
|
121
|
+
## Configuration
|
|
122
|
+
|
|
123
|
+
Settings live in `~/.orimem/config.toml`. Change them with `orimem config set <key> <value>` or ask your agent, which uses the `config_set` tool.
|
|
124
|
+
|
|
125
|
+
| Key | Default | Meaning |
|
|
126
|
+
|---|---|---|
|
|
127
|
+
| `vault` | `~/.orimem/vault` | Obsidian vault path |
|
|
128
|
+
| `capture.enabled` | `true` | Capture conversations into the vault |
|
|
129
|
+
| `capture.sources` | `["claude","codex","dsh"]` | Which agents to capture |
|
|
130
|
+
| `capture.since` | install time | Older sessions are not back-filled (`orimem capture --backfill` does it on request) |
|
|
131
|
+
| `capture.exclude_cwd` | `[]` | Never capture sessions started in these folders |
|
|
132
|
+
| `daily_summary.enabled` | `"ask"` | `"ask"` = ask once, off until answered · `true` · `false` |
|
|
133
|
+
| `daily_summary.time` | `"22:30"` | When the daily job runs |
|
|
134
|
+
| `daily_summary.agent` | auto | `claude` or `codex` runs the summary |
|
|
135
|
+
| `recall.hf_endpoint` | auto | Model download endpoint (falls back to hf-mirror.com) |
|
|
136
|
+
|
|
137
|
+
## Reference
|
|
138
|
+
|
|
139
|
+
<details>
|
|
140
|
+
<summary><b>CLI</b></summary>
|
|
141
|
+
|
|
142
|
+
| Command | What it does |
|
|
143
|
+
|---|---|
|
|
144
|
+
| `orimem init [-y]` | First-time setup |
|
|
145
|
+
| `orimem install / uninstall [--agent claude,codex,dsh] [--dsh-profile NAME]` | Connect or disconnect agents |
|
|
146
|
+
| `orimem doctor` | Check config, model, embedder, dependencies and agents |
|
|
147
|
+
| `orimem capture [--source X] [--session ID] [--backfill]` | Capture now |
|
|
148
|
+
| `orimem sync` · `orimem recall "query" --scope "10 项目"` | Index / search from the shell |
|
|
149
|
+
| `orimem daily status / run [--dry-run] / apply` | Daily summary |
|
|
150
|
+
| `orimem model status / download [--torch] / import DIR` | Embedding model |
|
|
151
|
+
| `orimem config get / set / path` | Settings |
|
|
152
|
+
| `orimem snippet` | MCP JSON + AGENTS.md snippet for other clients |
|
|
153
|
+
| `orimem --home DIR …` | Use another data directory (written into generated agent commands) |
|
|
154
|
+
</details>
|
|
155
|
+
|
|
156
|
+
<details>
|
|
157
|
+
<summary><b>MCP tools</b></summary>
|
|
158
|
+
|
|
159
|
+
| Tool | Purpose |
|
|
160
|
+
|---|---|
|
|
161
|
+
| `task_context` | Bind the current folder/task to a project entry |
|
|
162
|
+
| `recall` | Hybrid search (snippets) |
|
|
163
|
+
| `evidence` | Exact source text for result ids |
|
|
164
|
+
| `list_sessions` · `get_session_messages` | Browse captured conversations |
|
|
165
|
+
| `prepare_record` · `commit_record` | Verified write-back |
|
|
166
|
+
| `initialize` | Create a project/course entry |
|
|
167
|
+
| `lifecycle` | Source reading, candidates, corrections |
|
|
168
|
+
| `health` · `sync` · `archive_history` | Maintenance |
|
|
169
|
+
| `config_get` · `config_set` · `capture_now` | Settings and capture |
|
|
170
|
+
</details>
|
|
171
|
+
|
|
172
|
+
<details>
|
|
173
|
+
<summary><b>Vault layout</b></summary>
|
|
174
|
+
|
|
175
|
+
```text
|
|
176
|
+
00 收件箱/ inbox and knowledge candidates
|
|
177
|
+
10 项目/ one page per project (status, next action)
|
|
178
|
+
15 课程/ courses
|
|
179
|
+
30 知识/ reusable knowledge
|
|
180
|
+
50 日记/ 每日/ (daily) · 每周/ (weekly)
|
|
181
|
+
90 归档/ archive; 90 归档/Codex佐证/ holds captured transcripts
|
|
182
|
+
95 AI 工作区/ agent work notes
|
|
183
|
+
```
|
|
184
|
+
|
|
185
|
+
Writes are limited to `layout.folders`. `.obsidian` and files outside the vault are never read or modified.
|
|
186
|
+
</details>
|
|
187
|
+
|
|
188
|
+
## FAQ
|
|
189
|
+
|
|
190
|
+
**Do I need a GPU or PyTorch?** No. Embeddings run on CPU through ONNX Runtime. `orimem[torch]` plus `orimem model download --torch` enables the optional PyTorch backend, which produces identical vectors.
|
|
191
|
+
|
|
192
|
+
**Can I use my existing vault?** Yes. Point `vault` at it. OriMem only adds the folders it needs and never rewrites existing notes outside the verified write-back flow.
|
|
193
|
+
|
|
194
|
+
**I'm in mainland China.** Both installers detect an unreachable PyPI and switch to the Tsinghua PyPI mirror, the npmmirror Python builds and hf-mirror.com.
|
|
195
|
+
|
|
196
|
+
**How do I turn capture off for a project?** `orimem config set capture.exclude_cwd '["D:/work/secret"]'`. To turn it off everywhere: `orimem config set capture.enabled false`.
|
|
197
|
+
|
|
198
|
+
**How do I remove everything?** `orimem uninstall`, then `uv tool uninstall orimem-mcp`. Your vault and `~/.orimem` stay until you delete them.
|
|
199
|
+
|
|
200
|
+
## Roadmap
|
|
201
|
+
|
|
202
|
+
- [ ] Configurable / English vault layout
|
|
203
|
+
- [ ] More agents (Cursor, Gemini CLI, OpenCode)
|
|
204
|
+
- [ ] Optional quantized ONNX model for low-memory machines
|
|
205
|
+
|
|
206
|
+
## Contributing
|
|
207
|
+
|
|
208
|
+
```bash
|
|
209
|
+
git clone https://github.com/Wen5555/OriMem && cd OriMem
|
|
210
|
+
```
|
|
211
|
+
|
|
212
|
+
```bash
|
|
213
|
+
uv venv && uv pip install -e ".[test]"
|
|
214
|
+
```
|
|
215
|
+
|
|
216
|
+
```bash
|
|
217
|
+
uv run pytest
|
|
218
|
+
```
|
|
219
|
+
|
|
220
|
+
A manual end-to-end checklist for each agent is in [docs/TESTING.md](docs/TESTING.md). Issues and pull requests are welcome.
|
|
221
|
+
|
|
222
|
+
## License
|
|
223
|
+
|
|
224
|
+
[MIT](LICENSE)
|
|
@@ -0,0 +1,57 @@
|
|
|
1
|
+
[build-system]
|
|
2
|
+
requires = ["setuptools>=68"]
|
|
3
|
+
build-backend = "setuptools.build_meta"
|
|
4
|
+
|
|
5
|
+
[project]
|
|
6
|
+
name = "orimem-mcp"
|
|
7
|
+
version = "0.1.1"
|
|
8
|
+
description = "Local Obsidian-vault memory for coding agents (Claude Code, Codex, DeepSeek Harness) over MCP: hybrid recall, verified write-back, conversation capture and daily summaries."
|
|
9
|
+
readme = "README.md"
|
|
10
|
+
requires-python = ">=3.11"
|
|
11
|
+
license = { text = "MIT" }
|
|
12
|
+
authors = [{ name = "Orion" }]
|
|
13
|
+
keywords = ["mcp", "obsidian", "memory", "claude-code", "codex", "deepseek", "agent", "rag"]
|
|
14
|
+
classifiers = [
|
|
15
|
+
"Development Status :: 3 - Alpha",
|
|
16
|
+
"Environment :: Console",
|
|
17
|
+
"Intended Audience :: Developers",
|
|
18
|
+
"License :: OSI Approved :: MIT License",
|
|
19
|
+
"Operating System :: Microsoft :: Windows",
|
|
20
|
+
"Operating System :: MacOS",
|
|
21
|
+
"Operating System :: POSIX :: Linux",
|
|
22
|
+
"Programming Language :: Python :: 3.11",
|
|
23
|
+
"Programming Language :: Python :: 3.12",
|
|
24
|
+
"Programming Language :: Python :: 3.13",
|
|
25
|
+
"Topic :: Software Development",
|
|
26
|
+
]
|
|
27
|
+
dependencies = [
|
|
28
|
+
"numpy>=1.24",
|
|
29
|
+
"jsonschema>=4",
|
|
30
|
+
"pyyaml>=6",
|
|
31
|
+
"watchdog>=3",
|
|
32
|
+
"requests>=2.28",
|
|
33
|
+
"zstandard>=0.21",
|
|
34
|
+
"onnxruntime>=1.17",
|
|
35
|
+
"tokenizers>=0.19",
|
|
36
|
+
]
|
|
37
|
+
|
|
38
|
+
[project.optional-dependencies]
|
|
39
|
+
torch = ["torch>=2.1", "transformers>=4.40"]
|
|
40
|
+
test = ["pytest>=7"]
|
|
41
|
+
|
|
42
|
+
[project.scripts]
|
|
43
|
+
orimem = "orimem.cli:main"
|
|
44
|
+
|
|
45
|
+
[project.urls]
|
|
46
|
+
Homepage = "https://github.com/Wen5555/OriMem"
|
|
47
|
+
Documentation = "https://github.com/Wen5555/OriMem#readme"
|
|
48
|
+
Issues = "https://github.com/Wen5555/OriMem/issues"
|
|
49
|
+
|
|
50
|
+
[tool.setuptools.packages.find]
|
|
51
|
+
where = ["src"]
|
|
52
|
+
|
|
53
|
+
[tool.setuptools.package-data]
|
|
54
|
+
orimem = ["skill/**/*", "templates/**/*"]
|
|
55
|
+
|
|
56
|
+
[tool.pytest.ini_options]
|
|
57
|
+
testpaths = ["tests"]
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
__version__ = "0.1.1"
|