qwerty-agent 2.0.0__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.
Files changed (10) hide show
  1. qwerty_agent-2.0.0/LICENSE +21 -0
  2. qwerty_agent-2.0.0/PKG-INFO +178 -0
  3. qwerty_agent-2.0.0/README.md +153 -0
  4. qwerty_agent-2.0.0/pyproject.toml +40 -0
  5. qwerty_agent-2.0.0/qwerty_agent.egg-info/PKG-INFO +178 -0
  6. qwerty_agent-2.0.0/qwerty_agent.egg-info/SOURCES.txt +8 -0
  7. qwerty_agent-2.0.0/qwerty_agent.egg-info/dependency_links.txt +1 -0
  8. qwerty_agent-2.0.0/qwerty_agent.egg-info/entry_points.txt +4 -0
  9. qwerty_agent-2.0.0/qwerty_agent.egg-info/top_level.txt +4 -0
  10. qwerty_agent-2.0.0/setup.cfg +4 -0
qwerty_agent-2.0.0/LICENSE ADDED
@@ -0,0 +1,21 @@
1
+ MIT License
2
+
3
+ Copyright (c) 2026 KasishStar
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.
qwerty_agent-2.0.0/PKG-INFO ADDED
@@ -0,0 +1,178 @@
1
+ Metadata-Version: 2.4
2
+ Name: qwerty-agent
3
+ Version: 2.0.0
4
+ Summary: Zero-dependency symbolic AI agent. Pure Python stdlib. Runs anywhere.
5
+ Author: KasishStar
6
+ License: MIT
7
+ Project-URL: Homepage, https://github.com/KasishStar/qwerty-agent
8
+ Project-URL: Source, https://github.com/KasishStar/qwerty-agent
9
+ Project-URL: Tracker, https://github.com/KasishStar/qwerty-agent/issues
10
+ Keywords: ai,agent,symbolic,automation,cli
11
+ Classifier: Development Status :: 4 - Beta
12
+ Classifier: Environment :: Console
13
+ Classifier: Intended Audience :: Developers
14
+ Classifier: License :: OSI Approved :: MIT License
15
+ Classifier: Programming Language :: Python :: 3
16
+ Classifier: Programming Language :: Python :: 3.10
17
+ Classifier: Programming Language :: Python :: 3.11
18
+ Classifier: Programming Language :: Python :: 3.12
19
+ Classifier: Topic :: Software Development :: Build Tools
20
+ Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
21
+ Requires-Python: >=3.10
22
+ Description-Content-Type: text/markdown
23
+ License-File: LICENSE
24
+ Dynamic: license-file
25
+
26
+ # Qwerty v2 — Zero-Dependency Symbolic AI Agent
27
+
28
+ **Pure Python stdlib. No numpy. No torch. No sklearn. 10MB total. Runs on a Raspberry Pi.**
29
+
30
+ Qwerty is an autonomous software agent that reads files, edits code, runs commands, searches the web, learns from every interaction, and composes multi-source answers — all with **zero external dependencies**. It's a symbolic AI: deterministic, constitution-bound, auditable, and completely offline-capable.
31
+
32
+ ## Why Qwerty?
33
+
34
+ | Qwerty | Other AI Agents |
35
+ |---|---|
36
+ | `pip install` → runs | Requires Python ML stack (numpy, torch, transformers, 800MB+) |
37
+ | Offline by default | Cloud API or local LLM (8GB+ RAM) |
38
+ | 10MB on disk | 2-20GB for local models |
39
+ | Constitution hardcoded at engine level | Safety is a prompt (jailbreakable) |
40
+ | Deterministic — same input, same output | Non-deterministic (sampling) |
41
+ | Runs on ARM (Raspberry Pi Zero) | Needs x86 GPU or cloud |
42
+
43
+ ## Features
44
+
45
+ - **14 tools**: read, write, edit, search, find, run, web search, wikipedia, learn, analyze errors, and more
46
+ - **Fuzzy NLP**: handles `"runn ls -la"`, `"plz tel me bout the kernul"`, `"idk wat uefi is"` — 50+ shorthand maps, Levenshtein, n-gram Jaccard, contraction expansion, character dedup
47
+ - **Knowledge base**: 8 curated domains + 127 Wikipedia articles covering programming, OS dev, networking, math, physics, chemistry, biology, technology, and general knowledge
48
+ - **Genius composition engine**: retrieves, scores, and assembles multi-sentence responses from 2500+ sentences across all knowledge files
49
+ - **On-demand learning**: `lookup <topic>` → Wikipedia API (free, caches for next time); web fallback → DuckDuckGo
50
+ - **Constitution safety**: 9 hardcoded rules block `rm -rf /`, credential leaks, file overwrites without confirmation, and other dangerous operations
51
+ - **Persistent memory**: every answer, every lookup, every fix is stored in `memory/learned.json` — learning compounds across sessions
52
+ - **Recursive solve loop**: react → think → explore → internet → report — 3 layers of fallback before giving up
53
+ - **Protocol**: JSON-RPC, MCP server, and CLI modes — plugs into any editor or toolchain
54
+ - **76 passing tests**, zero external dependencies
55
+
56
+ ## Quick Start
57
+
58
+ ```bash
59
+ git clone https://github.com/KasishStar/qwerty-agent
60
+ cd qwerty-agent
61
+ python agent.py
62
+ ```
63
+
64
+ Or ask a question directly:
65
+
66
+ ```bash
67
+ python agent.py "what is a kernel"
68
+ ```
69
+
70
+ ## Interactive Mode
71
+
72
+ ```bash
73
+ python agent.py
74
+ ```
75
+
76
+ ```
77
+ qwerty> what is UEFI boot
78
+ qwerty> lookup quantum computing
79
+ qwerty> list files
80
+ qwerty> read file agent.py
81
+ qwerty> learn that the answer is 42
82
+ qwerty> exit
83
+ ```
84
+
85
+ ## Interface (with history and slash commands)
86
+
87
+ ```bash
88
+ python interface.py
89
+ ```
90
+
91
+ ## JSON-RPC Protocol
92
+
93
+ ```bash
94
+ python protocol.py --process "what is a kernel"
95
+ python protocol.py --json "what is a kernel"
96
+ python protocol.py --schema
97
+ python protocol.py --mcp
98
+ ```
99
+
100
+ ## Architecture
101
+
102
+ ```
103
+ User Input → Normalize → Classify Intent → Constitution Check → Execute → Result
104
+ ↓
105
+ Genius Composition Engine
106
+ (2500+ sentences, 8+ domains)
107
+ ↓
108
+ Recursive Fallback Loop
109
+ react → routine → explore → internet → report
110
+ ↓
111
+ Persistent Memory (learned.json)
112
+ ```
113
+
114
+ All layers are checked against the constitution before execution. No layer can run `rm -rf /`, overwrite user files without confirmation, or leak credentials.
115
+
116
+ ## How It Works
117
+
118
+ Qwerty is not an LLM. It's a **symbolic AI** — every decision is deterministic if/then logic. Its "intelligence" comes from:
119
+
120
+ 1. **Broad knowledge** (Wikipedia articles + curated domains)
121
+ 2. **Fuzzy pattern matching** (Levenshtein, n-gram Jaccard, keyword overlap)
122
+ 3. **Multi-source composition** (Genius scores and assembles sentences)
123
+ 4. **On-demand learning** (fetches and stores new knowledge at runtime)
124
+ 5. **Compounding memory** (every success is saved for next session)
125
+
126
+ ## What Qwerty Can Do
127
+
128
+ | Task | Example |
129
+ |---|---|
130
+ | Answer knowledge questions | `"what is a microkernel"` |
131
+ | Read/write files | `"read file agent.py"` |
132
+ | Edit code | `"edit line 10 of test.py to say hello"` |
133
+ | Run commands | `"run cargo build"` |
134
+ | Search code | `"search for TODO in src/"` |
135
+ | Learn new facts | `"learn that X that Y"` |
136
+ | Wikipedia lookup | `"lookup TCP protocol"` |
137
+ | Web search | `"search web for rust uefi tutorial"` |
138
+ | Analyze errors | Automatic on command failure |
139
+ | Remember across sessions | All in `memory/learned.json` |
140
+
141
+ ## Requirements
142
+
143
+ - Python 3.10+
144
+ - No pip packages needed
145
+ - Linux, macOS, or Windows
146
+ - ARM (Raspberry Pi) supported
147
+ - Internet optional (for web search and Wikipedia)
148
+
149
+ ## Project Structure
150
+
151
+ ```
152
+ qwerty-agent/
153
+ ├── agent.py # Main engine: constitution, tools, NLP, solve loop
154
+ ├── genius.py # Response composition engine
155
+ ├── interface.py # Interactive REPL with history
156
+ ├── protocol.py # JSON-RPC / MCP / CLI modes
157
+ ├── constitution.json # 9 safety rules
158
+ ├── knowledge/ # 9 JSON knowledge files (8 curated + Wikipedia)
159
+ ├── memory/ # Persistent learning (auto-generated)
160
+ ├── workflows/ # Pre-defined multi-step workflows
161
+ ├── knowledge_importer.py # Wikipedia article downloader
162
+ ├── test_all.py # 76 comprehensive tests
163
+ └── pyproject.toml # Build configuration
164
+ ```
165
+
166
+ ## Tests
167
+
168
+ ```bash
169
+ python test_all.py
170
+ ```
171
+
172
+ ## License
173
+
174
+ MIT — see [LICENSE](LICENSE).
175
+
176
+ ---
177
+
178
+ **Created by KasishStar.** Pure symbolic AI. No ML dependencies. Runs anywhere.
qwerty_agent-2.0.0/README.md ADDED
@@ -0,0 +1,153 @@
1
+ # Qwerty v2 — Zero-Dependency Symbolic AI Agent
2
+
3
+ **Pure Python stdlib. No numpy. No torch. No sklearn. 10MB total. Runs on a Raspberry Pi.**
4
+
5
+ Qwerty is an autonomous software agent that reads files, edits code, runs commands, searches the web, learns from every interaction, and composes multi-source answers — all with **zero external dependencies**. It's a symbolic AI: deterministic, constitution-bound, auditable, and completely offline-capable.
6
+
7
+ ## Why Qwerty?
8
+
9
+ | Qwerty | Other AI Agents |
10
+ |---|---|
11
+ | `pip install` → runs | Requires Python ML stack (numpy, torch, transformers, 800MB+) |
12
+ | Offline by default | Cloud API or local LLM (8GB+ RAM) |
13
+ | 10MB on disk | 2-20GB for local models |
14
+ | Constitution hardcoded at engine level | Safety is a prompt (jailbreakable) |
15
+ | Deterministic — same input, same output | Non-deterministic (sampling) |
16
+ | Runs on ARM (Raspberry Pi Zero) | Needs x86 GPU or cloud |
17
+
18
+ ## Features
19
+
20
+ - **14 tools**: read, write, edit, search, find, run, web search, wikipedia, learn, analyze errors, and more
21
+ - **Fuzzy NLP**: handles `"runn ls -la"`, `"plz tel me bout the kernul"`, `"idk wat uefi is"` — 50+ shorthand maps, Levenshtein, n-gram Jaccard, contraction expansion, character dedup
22
+ - **Knowledge base**: 8 curated domains + 127 Wikipedia articles covering programming, OS dev, networking, math, physics, chemistry, biology, technology, and general knowledge
23
+ - **Genius composition engine**: retrieves, scores, and assembles multi-sentence responses from 2500+ sentences across all knowledge files
24
+ - **On-demand learning**: `lookup <topic>` → Wikipedia API (free, caches for next time); web fallback → DuckDuckGo
25
+ - **Constitution safety**: 9 hardcoded rules block `rm -rf /`, credential leaks, file overwrites without confirmation, and other dangerous operations
26
+ - **Persistent memory**: every answer, every lookup, every fix is stored in `memory/learned.json` — learning compounds across sessions
27
+ - **Recursive solve loop**: react → think → explore → internet → report — 3 layers of fallback before giving up
28
+ - **Protocol**: JSON-RPC, MCP server, and CLI modes — plugs into any editor or toolchain
29
+ - **76 passing tests**, zero external dependencies
30
+
31
+ ## Quick Start
32
+
33
+ ```bash
34
+ git clone https://github.com/KasishStar/qwerty-agent
35
+ cd qwerty-agent
36
+ python agent.py
37
+ ```
38
+
39
+ Or ask a question directly:
40
+
41
+ ```bash
42
+ python agent.py "what is a kernel"
43
+ ```
44
+
45
+ ## Interactive Mode
46
+
47
+ ```bash
48
+ python agent.py
49
+ ```
50
+
51
+ ```
52
+ qwerty> what is UEFI boot
53
+ qwerty> lookup quantum computing
54
+ qwerty> list files
55
+ qwerty> read file agent.py
56
+ qwerty> learn that the answer is 42
57
+ qwerty> exit
58
+ ```
59
+
60
+ ## Interface (with history and slash commands)
61
+
62
+ ```bash
63
+ python interface.py
64
+ ```
65
+
66
+ ## JSON-RPC Protocol
67
+
68
+ ```bash
69
+ python protocol.py --process "what is a kernel"
70
+ python protocol.py --json "what is a kernel"
71
+ python protocol.py --schema
72
+ python protocol.py --mcp
73
+ ```
74
+
75
+ ## Architecture
76
+
77
+ ```
78
+ User Input → Normalize → Classify Intent → Constitution Check → Execute → Result
79
+ ↓
80
+ Genius Composition Engine
81
+ (2500+ sentences, 8+ domains)
82
+ ↓
83
+ Recursive Fallback Loop
84
+ react → routine → explore → internet → report
85
+ ↓
86
+ Persistent Memory (learned.json)
87
+ ```
88
+
89
+ All layers are checked against the constitution before execution. No layer can run `rm -rf /`, overwrite user files without confirmation, or leak credentials.
90
+
91
+ ## How It Works
92
+
93
+ Qwerty is not an LLM. It's a **symbolic AI** — every decision is deterministic if/then logic. Its "intelligence" comes from:
94
+
95
+ 1. **Broad knowledge** (Wikipedia articles + curated domains)
96
+ 2. **Fuzzy pattern matching** (Levenshtein, n-gram Jaccard, keyword overlap)
97
+ 3. **Multi-source composition** (Genius scores and assembles sentences)
98
+ 4. **On-demand learning** (fetches and stores new knowledge at runtime)
99
+ 5. **Compounding memory** (every success is saved for next session)
100
+
101
+ ## What Qwerty Can Do
102
+
103
+ | Task | Example |
104
+ |---|---|
105
+ | Answer knowledge questions | `"what is a microkernel"` |
106
+ | Read/write files | `"read file agent.py"` |
107
+ | Edit code | `"edit line 10 of test.py to say hello"` |
108
+ | Run commands | `"run cargo build"` |
109
+ | Search code | `"search for TODO in src/"` |
110
+ | Learn new facts | `"learn that X that Y"` |
111
+ | Wikipedia lookup | `"lookup TCP protocol"` |
112
+ | Web search | `"search web for rust uefi tutorial"` |
113
+ | Analyze errors | Automatic on command failure |
114
+ | Remember across sessions | All in `memory/learned.json` |
115
+
116
+ ## Requirements
117
+
118
+ - Python 3.10+
119
+ - No pip packages needed
120
+ - Linux, macOS, or Windows
121
+ - ARM (Raspberry Pi) supported
122
+ - Internet optional (for web search and Wikipedia)
123
+
124
+ ## Project Structure
125
+
126
+ ```
127
+ qwerty-agent/
128
+ ├── agent.py # Main engine: constitution, tools, NLP, solve loop
129
+ ├── genius.py # Response composition engine
130
+ ├── interface.py # Interactive REPL with history
131
+ ├── protocol.py # JSON-RPC / MCP / CLI modes
132
+ ├── constitution.json # 9 safety rules
133
+ ├── knowledge/ # 9 JSON knowledge files (8 curated + Wikipedia)
134
+ ├── memory/ # Persistent learning (auto-generated)
135
+ ├── workflows/ # Pre-defined multi-step workflows
136
+ ├── knowledge_importer.py # Wikipedia article downloader
137
+ ├── test_all.py # 76 comprehensive tests
138
+ └── pyproject.toml # Build configuration
139
+ ```
140
+
141
+ ## Tests
142
+
143
+ ```bash
144
+ python test_all.py
145
+ ```
146
+
147
+ ## License
148
+
149
+ MIT — see [LICENSE](LICENSE).
150
+
151
+ ---
152
+
153
+ **Created by KasishStar.** Pure symbolic AI. No ML dependencies. Runs anywhere.
qwerty_agent-2.0.0/pyproject.toml ADDED
@@ -0,0 +1,40 @@
1
+ [build-system]
2
+ requires = ["setuptools>=64"]
3
+ build-backend = "setuptools.build_meta"
4
+
5
+ [project]
6
+ name = "qwerty-agent"
7
+ version = "2.0.0"
8
+ description = "Zero-dependency symbolic AI agent. Pure Python stdlib. Runs anywhere."
9
+ readme = "README.md"
10
+ license = {text = "MIT"}
11
+ authors = [
12
+ {name = "KasishStar"}
13
+ ]
14
+ requires-python = ">=3.10"
15
+ keywords = ["ai", "agent", "symbolic", "automation", "cli"]
16
+ classifiers = [
17
+ "Development Status :: 4 - Beta",
18
+ "Environment :: Console",
19
+ "Intended Audience :: Developers",
20
+ "License :: OSI Approved :: MIT License",
21
+ "Programming Language :: Python :: 3",
22
+ "Programming Language :: Python :: 3.10",
23
+ "Programming Language :: Python :: 3.11",
24
+ "Programming Language :: Python :: 3.12",
25
+ "Topic :: Software Development :: Build Tools",
26
+ "Topic :: Scientific/Engineering :: Artificial Intelligence",
27
+ ]
28
+
29
+ [project.urls]
30
+ Homepage = "https://github.com/KasishStar/qwerty-agent"
31
+ Source = "https://github.com/KasishStar/qwerty-agent"
32
+ Tracker = "https://github.com/KasishStar/qwerty-agent/issues"
33
+
34
+ [project.scripts]
35
+ qwerty = "agent:main"
36
+ qwerty-json = "protocol:main"
37
+ qwerty-interface = "interface:main"
38
+
39
+ [tool.setuptools.packages.find]
40
+ include = ["*"]
qwerty_agent-2.0.0/qwerty_agent.egg-info/PKG-INFO ADDED
@@ -0,0 +1,178 @@
1
+ Metadata-Version: 2.4
2
+ Name: qwerty-agent
3
+ Version: 2.0.0
4
+ Summary: Zero-dependency symbolic AI agent. Pure Python stdlib. Runs anywhere.
5
+ Author: KasishStar
6
+ License: MIT
7
+ Project-URL: Homepage, https://github.com/KasishStar/qwerty-agent
8
+ Project-URL: Source, https://github.com/KasishStar/qwerty-agent
9
+ Project-URL: Tracker, https://github.com/KasishStar/qwerty-agent/issues
10
+ Keywords: ai,agent,symbolic,automation,cli
11
+ Classifier: Development Status :: 4 - Beta
12
+ Classifier: Environment :: Console
13
+ Classifier: Intended Audience :: Developers
14
+ Classifier: License :: OSI Approved :: MIT License
15
+ Classifier: Programming Language :: Python :: 3
16
+ Classifier: Programming Language :: Python :: 3.10
17
+ Classifier: Programming Language :: Python :: 3.11
18
+ Classifier: Programming Language :: Python :: 3.12
19
+ Classifier: Topic :: Software Development :: Build Tools
20
+ Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
21
+ Requires-Python: >=3.10
22
+ Description-Content-Type: text/markdown
23
+ License-File: LICENSE
24
+ Dynamic: license-file
25
+
26
+ # Qwerty v2 — Zero-Dependency Symbolic AI Agent
27
+
28
+ **Pure Python stdlib. No numpy. No torch. No sklearn. 10MB total. Runs on a Raspberry Pi.**
29
+
30
+ Qwerty is an autonomous software agent that reads files, edits code, runs commands, searches the web, learns from every interaction, and composes multi-source answers — all with **zero external dependencies**. It's a symbolic AI: deterministic, constitution-bound, auditable, and completely offline-capable.
31
+
32
+ ## Why Qwerty?
33
+
34
+ | Qwerty | Other AI Agents |
35
+ |---|---|
36
+ | `pip install` → runs | Requires Python ML stack (numpy, torch, transformers, 800MB+) |
37
+ | Offline by default | Cloud API or local LLM (8GB+ RAM) |
38
+ | 10MB on disk | 2-20GB for local models |
39
+ | Constitution hardcoded at engine level | Safety is a prompt (jailbreakable) |
40
+ | Deterministic — same input, same output | Non-deterministic (sampling) |
41
+ | Runs on ARM (Raspberry Pi Zero) | Needs x86 GPU or cloud |
42
+
43
+ ## Features
44
+
45
+ - **14 tools**: read, write, edit, search, find, run, web search, wikipedia, learn, analyze errors, and more
46
+ - **Fuzzy NLP**: handles `"runn ls -la"`, `"plz tel me bout the kernul"`, `"idk wat uefi is"` — 50+ shorthand maps, Levenshtein, n-gram Jaccard, contraction expansion, character dedup
47
+ - **Knowledge base**: 8 curated domains + 127 Wikipedia articles covering programming, OS dev, networking, math, physics, chemistry, biology, technology, and general knowledge
48
+ - **Genius composition engine**: retrieves, scores, and assembles multi-sentence responses from 2500+ sentences across all knowledge files
49
+ - **On-demand learning**: `lookup <topic>` → Wikipedia API (free, caches for next time); web fallback → DuckDuckGo
50
+ - **Constitution safety**: 9 hardcoded rules block `rm -rf /`, credential leaks, file overwrites without confirmation, and other dangerous operations
51
+ - **Persistent memory**: every answer, every lookup, every fix is stored in `memory/learned.json` — learning compounds across sessions
52
+ - **Recursive solve loop**: react → think → explore → internet → report — 3 layers of fallback before giving up
53
+ - **Protocol**: JSON-RPC, MCP server, and CLI modes — plugs into any editor or toolchain
54
+ - **76 passing tests**, zero external dependencies
55
+
56
+ ## Quick Start
57
+
58
+ ```bash
59
+ git clone https://github.com/KasishStar/qwerty-agent
60
+ cd qwerty-agent
61
+ python agent.py
62
+ ```
63
+
64
+ Or ask a question directly:
65
+
66
+ ```bash
67
+ python agent.py "what is a kernel"
68
+ ```
69
+
70
+ ## Interactive Mode
71
+
72
+ ```bash
73
+ python agent.py
74
+ ```
75
+
76
+ ```
77
+ qwerty> what is UEFI boot
78
+ qwerty> lookup quantum computing
79
+ qwerty> list files
80
+ qwerty> read file agent.py
81
+ qwerty> learn that the answer is 42
82
+ qwerty> exit
83
+ ```
84
+
85
+ ## Interface (with history and slash commands)
86
+
87
+ ```bash
88
+ python interface.py
89
+ ```
90
+
91
+ ## JSON-RPC Protocol
92
+
93
+ ```bash
94
+ python protocol.py --process "what is a kernel"
95
+ python protocol.py --json "what is a kernel"
96
+ python protocol.py --schema
97
+ python protocol.py --mcp
98
+ ```
99
+
100
+ ## Architecture
101
+
102
+ ```
103
+ User Input → Normalize → Classify Intent → Constitution Check → Execute → Result
104
+ ↓
105
+ Genius Composition Engine
106
+ (2500+ sentences, 8+ domains)
107
+ ↓
108
+ Recursive Fallback Loop
109
+ react → routine → explore → internet → report
110
+ ↓
111
+ Persistent Memory (learned.json)
112
+ ```
113
+
114
+ All layers are checked against the constitution before execution. No layer can run `rm -rf /`, overwrite user files without confirmation, or leak credentials.
115
+
116
+ ## How It Works
117
+
118
+ Qwerty is not an LLM. It's a **symbolic AI** — every decision is deterministic if/then logic. Its "intelligence" comes from:
119
+
120
+ 1. **Broad knowledge** (Wikipedia articles + curated domains)
121
+ 2. **Fuzzy pattern matching** (Levenshtein, n-gram Jaccard, keyword overlap)
122
+ 3. **Multi-source composition** (Genius scores and assembles sentences)
123
+ 4. **On-demand learning** (fetches and stores new knowledge at runtime)
124
+ 5. **Compounding memory** (every success is saved for next session)
125
+
126
+ ## What Qwerty Can Do
127
+
128
+ | Task | Example |
129
+ |---|---|
130
+ | Answer knowledge questions | `"what is a microkernel"` |
131
+ | Read/write files | `"read file agent.py"` |
132
+ | Edit code | `"edit line 10 of test.py to say hello"` |
133
+ | Run commands | `"run cargo build"` |
134
+ | Search code | `"search for TODO in src/"` |
135
+ | Learn new facts | `"learn that X that Y"` |
136
+ | Wikipedia lookup | `"lookup TCP protocol"` |
137
+ | Web search | `"search web for rust uefi tutorial"` |
138
+ | Analyze errors | Automatic on command failure |
139
+ | Remember across sessions | All in `memory/learned.json` |
140
+
141
+ ## Requirements
142
+
143
+ - Python 3.10+
144
+ - No pip packages needed
145
+ - Linux, macOS, or Windows
146
+ - ARM (Raspberry Pi) supported
147
+ - Internet optional (for web search and Wikipedia)
148
+
149
+ ## Project Structure
150
+
151
+ ```
152
+ qwerty-agent/
153
+ ├── agent.py # Main engine: constitution, tools, NLP, solve loop
154
+ ├── genius.py # Response composition engine
155
+ ├── interface.py # Interactive REPL with history
156
+ ├── protocol.py # JSON-RPC / MCP / CLI modes
157
+ ├── constitution.json # 9 safety rules
158
+ ├── knowledge/ # 9 JSON knowledge files (8 curated + Wikipedia)
159
+ ├── memory/ # Persistent learning (auto-generated)
160
+ ├── workflows/ # Pre-defined multi-step workflows
161
+ ├── knowledge_importer.py # Wikipedia article downloader
162
+ ├── test_all.py # 76 comprehensive tests
163
+ └── pyproject.toml # Build configuration
164
+ ```
165
+
166
+ ## Tests
167
+
168
+ ```bash
169
+ python test_all.py
170
+ ```
171
+
172
+ ## License
173
+
174
+ MIT — see [LICENSE](LICENSE).
175
+
176
+ ---
177
+
178
+ **Created by KasishStar.** Pure symbolic AI. No ML dependencies. Runs anywhere.
qwerty_agent-2.0.0/qwerty_agent.egg-info/SOURCES.txt ADDED
@@ -0,0 +1,8 @@
1
+ LICENSE
2
+ README.md
3
+ pyproject.toml
4
+ qwerty_agent.egg-info/PKG-INFO
5
+ qwerty_agent.egg-info/SOURCES.txt
6
+ qwerty_agent.egg-info/dependency_links.txt
7
+ qwerty_agent.egg-info/entry_points.txt
8
+ qwerty_agent.egg-info/top_level.txt
qwerty_agent-2.0.0/qwerty_agent.egg-info/entry_points.txt ADDED
@@ -0,0 +1,4 @@
1
+ [console_scripts]
2
+ qwerty = agent:main
3
+ qwerty-interface = interface:main
4
+ qwerty-json = protocol:main
qwerty_agent-2.0.0/qwerty_agent.egg-info/top_level.txt ADDED
@@ -0,0 +1,4 @@
1
+ dist
2
+ knowledge
3
+ memory
4
+ workflows
qwerty_agent-2.0.0/setup.cfg ADDED
@@ -0,0 +1,4 @@
1
+ [egg_info]
2
+ tag_build =
3
+ tag_date = 0
4
+