moot-memory 1.0.24__py3-none-any.whl
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.
|
@@ -0,0 +1,119 @@
|
|
|
1
|
+
Metadata-Version: 2.4
|
|
2
|
+
Name: moot-memory
|
|
3
|
+
Version: 1.0.24
|
|
4
|
+
Summary: Drop-in /memories backend with governance for the Anthropic memory_20250818 tool
|
|
5
|
+
Author: Codedaptive
|
|
6
|
+
License: MIT
|
|
7
|
+
Project-URL: Homepage, https://github.com/codedaptive/mootx01-ce
|
|
8
|
+
Project-URL: Repository, https://github.com/codedaptive/mootx01-ce
|
|
9
|
+
Project-URL: Documentation, https://github.com/codedaptive/mootx01-ce/tree/stable/1.0.x/apps/moot-memory-adapter
|
|
10
|
+
Keywords: anthropic,claude,memory,mcp,mootx01,ai-memory
|
|
11
|
+
Classifier: Development Status :: 4 - Beta
|
|
12
|
+
Classifier: Intended Audience :: Developers
|
|
13
|
+
Classifier: License :: OSI Approved :: MIT License
|
|
14
|
+
Classifier: Programming Language :: Python :: 3
|
|
15
|
+
Classifier: Programming Language :: Python :: 3.10
|
|
16
|
+
Classifier: Programming Language :: Python :: 3.11
|
|
17
|
+
Classifier: Programming Language :: Python :: 3.12
|
|
18
|
+
Classifier: Programming Language :: Python :: 3.13
|
|
19
|
+
Classifier: Topic :: Software Development :: Libraries
|
|
20
|
+
Requires-Python: >=3.10
|
|
21
|
+
Description-Content-Type: text/markdown
|
|
22
|
+
Provides-Extra: anthropic
|
|
23
|
+
Requires-Dist: anthropic>=0.40.0; extra == "anthropic"
|
|
24
|
+
|
|
25
|
+
# MOOTx01 Memory Adapter
|
|
26
|
+
|
|
27
|
+
Drop-in `/memories` backend with governance for the Anthropic
|
|
28
|
+
`memory_20250818` tool.
|
|
29
|
+
|
|
30
|
+
## What it does
|
|
31
|
+
|
|
32
|
+
Any Claude 4+ / Fable 5 agent that thinks it is reading and writing
|
|
33
|
+
`/memories` files is actually reading and writing **governed MOOTx01
|
|
34
|
+
drawers** — with dedup, provenance, sensitivity floor, confirmation
|
|
35
|
+
state, and a full audit trail included.
|
|
36
|
+
|
|
37
|
+
## The differentiator
|
|
38
|
+
|
|
39
|
+
Every model-written memory lands as **unconfirmed** with derived trust.
|
|
40
|
+
A startup scan or audit pass can quarantine suspect lessons using the
|
|
41
|
+
estate's contained filter and anomaly lenses. This is the poisoning
|
|
42
|
+
defense the ecosystem currently lacks — the memory tool contract gives
|
|
43
|
+
models write access; MOOTx01 gives you the governance to trust (or not
|
|
44
|
+
trust) what they wrote.
|
|
45
|
+
|
|
46
|
+
## Two surfaces
|
|
47
|
+
|
|
48
|
+
### 1. MCP tool (built into mootx01)
|
|
49
|
+
|
|
50
|
+
The mootx01 daemon registers a `memory` tool on its MCP surface that
|
|
51
|
+
matches Anthropic's contract exactly. Claude Code and Claude Desktop
|
|
52
|
+
users already connected to mootx01 as an MCP server get the memory tool
|
|
53
|
+
for free — no adapter needed.
|
|
54
|
+
|
|
55
|
+
### 2. Python SDK handler (this package)
|
|
56
|
+
|
|
57
|
+
For Messages API users who run the tool-use loop themselves:
|
|
58
|
+
|
|
59
|
+
```python
|
|
60
|
+
from moot_memory import MootMemoryTool
|
|
61
|
+
import anthropic
|
|
62
|
+
|
|
63
|
+
client = anthropic.Anthropic()
|
|
64
|
+
memory = MootMemoryTool(base_url="http://127.0.0.1:4242")
|
|
65
|
+
|
|
66
|
+
runner = client.beta.messages.tool_runner(
|
|
67
|
+
model="claude-opus-4-8",
|
|
68
|
+
max_tokens=1024,
|
|
69
|
+
messages=[{"role": "user", "content": "Remember that Acme prefers email."}],
|
|
70
|
+
tools=[memory],
|
|
71
|
+
)
|
|
72
|
+
final = runner.until_done()
|
|
73
|
+
```
|
|
74
|
+
|
|
75
|
+
Or standalone without the SDK:
|
|
76
|
+
|
|
77
|
+
```python
|
|
78
|
+
from moot_memory import MootMemoryHandler
|
|
79
|
+
|
|
80
|
+
handler = MootMemoryHandler(base_url="http://127.0.0.1:4242")
|
|
81
|
+
result = handler.execute({"command": "view", "path": "/memories"})
|
|
82
|
+
```
|
|
83
|
+
|
|
84
|
+
## Op mapping
|
|
85
|
+
|
|
86
|
+
| memory_20250818 | Estate behavior |
|
|
87
|
+
|---|---|
|
|
88
|
+
| view (dir) | enumerate drawers in wing="memories" |
|
|
89
|
+
| view (file) | drawer content with line numbers |
|
|
90
|
+
| create | capture as unconfirmed drawer |
|
|
91
|
+
| str_replace | supersede: new content, old withdrawn (full lineage) |
|
|
92
|
+
| insert | supersede: content with insertion applied |
|
|
93
|
+
| delete | soft withdrawal (never hard-erase from model ops) |
|
|
94
|
+
| rename | capture at new location, withdraw old |
|
|
95
|
+
|
|
96
|
+
## Security
|
|
97
|
+
|
|
98
|
+
- **Path traversal protection**: all paths validated against `/memories`
|
|
99
|
+
- **Model writes land unconfirmed**: poisoning quarantine seam
|
|
100
|
+
- **Deletes are soft withdrawals**: reversible, audit trail preserved
|
|
101
|
+
- **Sensitivity floor**: adapter wing is Normal
|
|
102
|
+
- **File size cap**: 100KB per file
|
|
103
|
+
- **No hidden files**: dotfiles rejected
|
|
104
|
+
|
|
105
|
+
## Tests
|
|
106
|
+
|
|
107
|
+
```bash
|
|
108
|
+
# Against the running daemon
|
|
109
|
+
python -m pytest apps/moot-memory-adapter/tests/
|
|
110
|
+
|
|
111
|
+
# The poisoning quarantine test
|
|
112
|
+
python -m pytest apps/moot-memory-adapter/tests/test_memory_adapter.py::TestPoisoningQuarantine -v
|
|
113
|
+
```
|
|
114
|
+
|
|
115
|
+
## Requirements
|
|
116
|
+
|
|
117
|
+
- Running mootx01 daemon (v1.0.22+)
|
|
118
|
+
- Python 3.10+
|
|
119
|
+
- `anthropic` SDK (optional, for the BetaAbstractMemoryTool subclass)
|
|
@@ -0,0 +1,4 @@
|
|
|
1
|
+
moot_memory-1.0.24.dist-info/METADATA,sha256=-X9XAWmqJeHCmS1jTPpBVywIo_BrZJDZVTOg3DpgaRA,4026
|
|
2
|
+
moot_memory-1.0.24.dist-info/WHEEL,sha256=K260EYznzXsJYBQGqmI8VTxEdiZYNvDZwW9cBh9-_MA,91
|
|
3
|
+
moot_memory-1.0.24.dist-info/top_level.txt,sha256=AbpHGcgLb-kRsJGnwFEktk7uzpZOCcBY74-YBdrKVGs,1
|
|
4
|
+
moot_memory-1.0.24.dist-info/RECORD,,
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
|