krenk 0.1.0

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.
package/LICENSE ADDED
@@ -0,0 +1,21 @@
1
+ MIT License
2
+
3
+ Copyright (c) 2026 Krishna Kant Singh
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.
package/README.md ADDED
@@ -0,0 +1,172 @@
1
+ <p align="center">
2
+ <img src="public/img/main.png" alt="Krenk CLI" width="700" />
3
+ </p>
4
+
5
+ # Krenk
6
+
7
+ **Like Claude Code, but with a team.**
8
+
9
+ Krenk is a multi-agent software engineering CLI that orchestrates an entire team of specialized AI agents to build your project -- from planning to deployment. Instead of one AI doing everything, Krenk assigns the right job to the right agent, just like a real engineering team.
10
+
11
+ ---
12
+
13
+ ## Why Krenk over plain Claude Code?
14
+
15
+ Claude Code is powerful, but it's one agent doing everything: planning, coding, testing, reviewing, documenting. That's like asking a single developer to be the architect, the frontend engineer, the QA lead, the security auditor, and the tech writer -- all at once, in one context window.
16
+
17
+ **Krenk fixes this.**
18
+
19
+ | | Claude Code | Krenk |
20
+ |---|---|---|
21
+ | Agents | 1 | Up to 11 specialized agents |
22
+ | Planning | Inline, ad-hoc | Dedicated Strategist creates a master plan with per-agent assignments |
23
+ | Architecture | Mixed with implementation | Dedicated Architect sets up skeleton before Builder writes features |
24
+ | Code review | You review manually | Sentinel agent reviews code automatically for bugs, security, and quality |
25
+ | Testing | You write tests or ask | Guardian agent writes and runs comprehensive test suites |
26
+ | Security | Hope for the best | Shield agent performs a dedicated security audit |
27
+ | Coordination | Single context window | Master Brain coordinates agents via shared memory |
28
+ | Supervision | None | Process Supervisor monitors memory, CPU, and kills hung agents |
29
+ | Parallelism | Sequential | Parallel builders for independent modules |
30
+
31
+ **The result:** Higher quality output, faster execution for large projects, and specialized attention to each concern -- planning, design, architecture, implementation, testing, review, security, documentation, and deployment all get dedicated focus.
32
+
33
+ ---
34
+
35
+ ## How it works
36
+
37
+ 1. **You describe what you want to build** -- plain English, typos and all
38
+ 2. **Krenk refines your prompt** -- using Claude to extract requirements, tech stack, and structure
39
+ 3. **The Strategist creates a master plan** -- breaking work into tasks and assigning them to agents
40
+ 4. **Agents execute in sequence** -- each one focused on its specialty, building on the previous agent's output
41
+ 5. **The Master Brain coordinates** -- sharing context, reviewing output, ordering redos if quality is low
42
+ 6. **The Supervisor watches everything** -- monitoring memory, CPU, and killing hung processes
43
+
44
+ ### The Team
45
+
46
+ | Agent | Role | What it does |
47
+ |---|---|---|
48
+ | **Analyst** | Business analysis | User stories, acceptance criteria, priorities |
49
+ | **Strategist** | Planning | Master plan with per-agent task assignments |
50
+ | **Pixel** | UI/UX design | Component hierarchy, layouts, styling specs |
51
+ | **Blueprint** | Architecture | System design, project skeleton, data models |
52
+ | **Builder** | Implementation | Writes the actual production code |
53
+ | **QA Lead** | Test planning | Test strategy, test cases, coverage matrix |
54
+ | **Guardian** | Testing | Writes and runs unit/integration tests |
55
+ | **Sentinel** | Code review | Reviews for bugs, performance, and style |
56
+ | **Shield** | Security audit | OWASP checks, dependency audit, auth review |
57
+ | **Scribe** | Documentation | README, API docs, inline comments |
58
+ | **DevOps** | Deployment | CI/CD, Docker, deployment configs |
59
+
60
+ Not every project needs all 11. Krenk lets you pick your team:
61
+
62
+ - **Full Team** -- all 11 agents
63
+ - **Engineering** -- plan, architect, code, test, review
64
+ - **QA Focused** -- plan, code, QA, test, review
65
+ - **Startup MVP** -- plan, architect, code, review, docs
66
+ - **Quick Build** -- plan and code only
67
+ - **Custom** -- pick your own agents
68
+
69
+ ---
70
+
71
+ ## Installation
72
+
73
+ ### Prerequisites
74
+
75
+ - Node.js >= 20
76
+ - [Claude Code CLI](https://docs.anthropic.com/en/docs/claude-code) installed and authenticated
77
+
78
+ ### Install from source
79
+
80
+ ```bash
81
+ git clone https://github.com/yourusername/krenk.git
82
+ cd krenk
83
+ npm install
84
+ npm run build
85
+ npm link
86
+ ```
87
+
88
+ ### Run
89
+
90
+ ```bash
91
+ krenk
92
+ ```
93
+
94
+ That's it. The interactive UI walks you through everything.
95
+
96
+ ---
97
+
98
+ ## Architecture
99
+
100
+ ```
101
+ krenk
102
+ |-- src/
103
+ | |-- orchestrator/
104
+ | | |-- engine.ts # Main pipeline: runs agents through stages
105
+ | | |-- brain.ts # Master Brain: coordinates agents, reviews, redos
106
+ | | |-- memory.ts # Shared memory: .krenk/shared/ files all agents read/write
107
+ | | |-- supervisor.ts # Process watchdog: memory, CPU, timeout monitoring
108
+ | | |-- plan-parser.ts # Parses strategist output into per-agent assignments
109
+ | | |-- scheduler.ts # Parallel execution with concurrency control
110
+ | | |-- context.ts # Accumulates output from previous agents
111
+ | | |-- workflow.ts # Stage definitions and ordering
112
+ | |-- agents/
113
+ | | |-- roles.ts # Agent definitions: prompts, tools, boundaries
114
+ | | |-- spawner.ts # Spawns Claude CLI processes with stream-json
115
+ | | |-- registry.ts # Tracks active agents, PIDs, output
116
+ | |-- ui/
117
+ | | |-- interactive.ts # Interactive CLI: team selection, prompt, refinement
118
+ | | |-- renderer.ts # Real-time spinner with stream-json activity
119
+ | | |-- theme.ts # Colors and gradients
120
+ | |-- config/
121
+ | | |-- defaults.ts # Default config: max turns, parallel agents
122
+ | | |-- loader.ts # Config file loader (cosmiconfig)
123
+ | |-- utils/
124
+ | |-- logger.ts # Logging
125
+ | |-- process.ts # Graceful shutdown
126
+ ```
127
+
128
+ ### Key design decisions
129
+
130
+ **Stream-JSON output** -- Agents use `--output-format stream-json` so the CLI shows real-time activity (which file is being read, which tool is being called) instead of a frozen spinner.
131
+
132
+ **Tool restrictions** -- Read-only agents (Strategist, Analyst, Sentinel) have `--disallowedTools Write,Edit,Bash` enforced at the CLI level so they cannot accidentally write code.
133
+
134
+ **Shared memory** -- All agents can read `.krenk/shared/` markdown files. The Master Brain writes directives to `brain.md`, agents write their progress. This lets later agents build on earlier agents' decisions without passing massive context strings.
135
+
136
+ **Process supervision** -- Every spawned Claude process is tracked. If an agent exceeds 512MB memory, runs longer than 10 minutes, or produces no output for 5 minutes, the Supervisor kills it and the Brain can retry.
137
+
138
+ ---
139
+
140
+ ## Configuration
141
+
142
+ Create a `.krenkrc.json`, `.krenkrc.yml`, or `krenk.config.js` in your project root:
143
+
144
+ ```json
145
+ {
146
+ "maxParallelAgents": 3,
147
+ "agents": {
148
+ "strategist": { "maxTurns": 30 },
149
+ "architect": { "maxTurns": 50 },
150
+ "builder": { "maxTurns": 100 },
151
+ "guardian": { "maxTurns": 50 }
152
+ }
153
+ }
154
+ ```
155
+
156
+ ---
157
+
158
+ ## Modes
159
+
160
+ **Autonomous** -- agents run without asking. Fastest path from prompt to project.
161
+
162
+ **Supervised** -- you approve each agent before it runs. You see what it will do and can skip or abort.
163
+
164
+ ---
165
+
166
+ ## License
167
+
168
+ MIT
169
+
170
+ ---
171
+
172
+ Built by Krishna with mass amount of coffee.
package/bin/krenk.js ADDED
@@ -0,0 +1,2 @@
1
+ #!/usr/bin/env node
2
+ import('../dist/index.js');