agentic-python-coder 2.2.1__tar.gz → 3.0.0__tar.gz

{agentic_python_coder-2.2.1 → agentic_python_coder-3.0.0}/.gitignore

@@ -152,8 +152,10 @@ cython_debug/
 uv.lock
 
 # Project specific
+tests/
 coder-examples/
 CLAUDE-archive.md
+EXPERTISE/
 PROCESS_NOTES.md
 conversation_log.json
 coder_output.log
@@ -171,9 +173,6 @@ PAPER/
 PAPER-ASP/
 ZEBRA/
 .mcp.json
-examples/cpmpy/cpmpy_v*.md
-examples/cpmpy/cpmpy.md.backup-*
-examples/clingo/clingo_v*.md
 
 # Test files and folders (root level only)
 /test-*/

{agentic_python_coder-2.2.1 → agentic_python_coder-3.0.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: agentic-python-coder
-Version: 2.2.1
+Version: 3.0.0
 Summary: A lightweight Python coding agent that writes, executes, and iterates on code through natural language instructions
 Author: Stefan Szeider
 License: Apache-2.0
@@ -14,25 +14,21 @@ Classifier: Programming Language :: Python :: 3.13
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Classifier: Topic :: Software Development :: Code Generators
 Requires-Python: <3.14,>=3.13
-Requires-Dist: ipykernel>=6.30.1
-Requires-Dist: jupyter-client>=8.6.3
-Requires-Dist: langchain-anthropic>=1.2.0
-Requires-Dist: langchain-core>=1.1.0
-Requires-Dist: langchain-experimental>=0.4.0
-Requires-Dist: langchain-openai>=1.1.0
-Requires-Dist: langgraph>=1.0.4
-Requires-Dist: mcp>=1.0.0
+Requires-Dist: ipykernel>=7.1.0
+Requires-Dist: jupyter-client>=8.8.0
+Requires-Dist: mcp>=1.26.0
+Requires-Dist: openai>=2.16.0
 Requires-Dist: python-dotenv>=1.2.1
 Requires-Dist: pyyaml>=6.0.3
-Requires-Dist: rich>=14.2.0
+Requires-Dist: rich>=14.3.1
 Provides-Extra: dev
-Requires-Dist: mypy>=1.19.0; extra == 'dev'
-Requires-Dist: ruff>=0.14.7; extra == 'dev'
+Requires-Dist: mypy>=1.19.1; extra == 'dev'
+Requires-Dist: ruff>=0.14.14; extra == 'dev'
 Provides-Extra: test
-Requires-Dist: pytest-asyncio>=1.2.0; extra == 'test'
+Requires-Dist: pytest-asyncio>=1.3.0; extra == 'test'
 Requires-Dist: pytest-cov>=7.0.0; extra == 'test'
 Requires-Dist: pytest-watch>=4.2.0; extra == 'test'
-Requires-Dist: pytest>=9.0.1; extra == 'test'
+Requires-Dist: pytest>=9.0.2; extra == 'test'
 Description-Content-Type: text/markdown
 
 # Agentic Python Coder
@@ -41,11 +37,10 @@ Description-Content-Type: text/markdown
 [![License: Apache 2.0](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
 [![MCP Compatible](https://img.shields.io/badge/MCP-Compatible-green.svg)](https://modelcontextprotocol.io/)
 [![UV](https://img.shields.io/badge/Packaged%20with-UV-purple)](https://github.com/astral-sh/uv)
-[![LangGraph](https://img.shields.io/badge/Built%20with-LangGraph-green)](https://github.com/langchain-ai/langgraph)
 
 This package provides two utilities for Python code execution:
 
-1. **coder** — An autonomous coding agent using the ReAct framework (CLI + Python library)
+1. **coder** — An autonomous coding agent using the ReAct pattern (CLI + Python library)
 2. **ipython_mcp** — An MCP server that gives any MCP-compatible client (Claude Desktop, etc.) Python execution capability
 
 Both share a persistent IPython kernel for stateful code execution.
@@ -156,8 +151,9 @@ coder --model opus45 "task"     # Claude Opus 4.5
 coder --model deepseek31 "task" # DeepSeek v3.1
 coder --model grok41 "task"     # X.AI Grok 4.1
 coder --model qwen3 "task"      # Qwen3 Coder
-coder --model gemini25 "task"   # Gemini Pro 2.5
-coder --model gpt5 "task"       # GPT-5
+coder --model gemini25 "task"     # Gemini Pro 2.5
+coder --model gemini3pro "task"  # Gemini 3 Pro Preview
+coder --model gpt52 "task"       # GPT-5.2
 
 # Custom model (JSON file)
 coder --model ./mymodel.json "task"
@@ -165,16 +161,17 @@ coder --model ./mymodel.json "task"
 
 ### Project Templates
 
-Domain-specific templates improve results:
+Domain-specific templates improve results. Bundled examples are available on GitHub at [`coder/src/agentic_python_coder/examples/`](coder/src/agentic_python_coder/examples/). Use `--init` to copy them locally:
 
 ```bash
-# Initialize example templates (creates coder-examples/ directory)
+# Copy all bundled examples to coder-examples/
 coder --init
 
-# Constraint programming with CPMpy
-coder --with cpmpy --project coder-examples/cpmpy/cpmpy.md "Solve 8-queens"
+# Or copy a specific template
+coder --init cpmpy
 
-# Answer Set Programming with Clingo
+# Then use with your task
+coder --with cpmpy --project coder-examples/cpmpy/cpmpy.md "Solve 8-queens"
 coder --with clingo --project coder-examples/clingo/clingo.md "Model bird flight"
 ```
 
@@ -262,7 +259,7 @@ from agentic_python_coder import get_openrouter_llm, list_available_models
 
 llm = get_openrouter_llm(model="sonnet45")
 print(list_available_models())
-# ['deepseek31', 'gemini25', 'gpt5', 'grok41', 'opus45', 'qwen3', 'sonnet45']
+# ['deepseek31', 'gemini25', 'gemini3pro', 'gpt52', 'grok41', 'opus45', 'qwen3', 'sonnet45']
 ```
 
 ---
@@ -291,17 +288,32 @@ Add to your MCP settings (e.g., `~/.claude/claude_desktop_config.json` or projec
 | Tool | Description |
 |------|-------------|
 | `python_exec` | Execute Python code. Auto-starts session if needed. Default 30s timeout. |
-| `python_reset` | Clear session state. Optionally install packages (e.g., `packages=["numpy", "pandas"]`). |
-| `python_status` | Check if session is active, Python version, installed packages, defined variables. |
+| `python_reset` | Create new kernel (no `kernel_id`) OR reset existing kernel (with `kernel_id`). Optionally install packages. |
+| `python_status` | Check session state: active flag, all active kernel IDs, Python version, packages, variables. |
 | `python_interrupt` | Send interrupt signal to stop long-running code. Session state is preserved. |
 
+### Multi-Agent Workflow
+
+For parallel agents, each agent gets its own kernel:
+
+```
+Agent A                              Agent B
+────────                             ────────
+python_reset() → kernel_id="aaa"     python_reset() → kernel_id="bbb"
+python_exec(kernel_id="aaa", ...)    python_exec(kernel_id="bbb", ...)
+python_exec(kernel_id="aaa", ...)    python_exec(kernel_id="bbb", ...)
+```
+
+Simple single-agent use: just call `python_exec()` — the default kernel auto-starts.
+
 ### Features
 
 - **Persistent state**: Variables, imports, and definitions persist across executions
-- **Auto-start**: Session starts automatically on first `python_exec`
+- **Auto-start**: Default session starts automatically on first `python_exec`
 - **Package installation**: Use `python_reset` with `packages` parameter to install dependencies
 - **Timeout handling**: Long-running code times out gracefully (session preserved)
 - **Interrupt support**: Stop runaway code without losing session state
+- **Multi-kernel**: Each `python_reset()` creates an isolated kernel for parallel agents
 
 ### Usage Tips
 

{agentic_python_coder-2.2.1 → agentic_python_coder-3.0.0}/README.md

@@ -4,11 +4,10 @@
 [![License: Apache 2.0](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
 [![MCP Compatible](https://img.shields.io/badge/MCP-Compatible-green.svg)](https://modelcontextprotocol.io/)
 [![UV](https://img.shields.io/badge/Packaged%20with-UV-purple)](https://github.com/astral-sh/uv)
-[![LangGraph](https://img.shields.io/badge/Built%20with-LangGraph-green)](https://github.com/langchain-ai/langgraph)
 
 This package provides two utilities for Python code execution:
 
-1. **coder** — An autonomous coding agent using the ReAct framework (CLI + Python library)
+1. **coder** — An autonomous coding agent using the ReAct pattern (CLI + Python library)
 2. **ipython_mcp** — An MCP server that gives any MCP-compatible client (Claude Desktop, etc.) Python execution capability
 
 Both share a persistent IPython kernel for stateful code execution.
@@ -119,8 +118,9 @@ coder --model opus45 "task"     # Claude Opus 4.5
 coder --model deepseek31 "task" # DeepSeek v3.1
 coder --model grok41 "task"     # X.AI Grok 4.1
 coder --model qwen3 "task"      # Qwen3 Coder
-coder --model gemini25 "task"   # Gemini Pro 2.5
-coder --model gpt5 "task"       # GPT-5
+coder --model gemini25 "task"     # Gemini Pro 2.5
+coder --model gemini3pro "task"  # Gemini 3 Pro Preview
+coder --model gpt52 "task"       # GPT-5.2
 
 # Custom model (JSON file)
 coder --model ./mymodel.json "task"
@@ -128,16 +128,17 @@ coder --model ./mymodel.json "task"
 
 ### Project Templates
 
-Domain-specific templates improve results:
+Domain-specific templates improve results. Bundled examples are available on GitHub at [`coder/src/agentic_python_coder/examples/`](coder/src/agentic_python_coder/examples/). Use `--init` to copy them locally:
 
 ```bash
-# Initialize example templates (creates coder-examples/ directory)
+# Copy all bundled examples to coder-examples/
 coder --init
 
-# Constraint programming with CPMpy
-coder --with cpmpy --project coder-examples/cpmpy/cpmpy.md "Solve 8-queens"
+# Or copy a specific template
+coder --init cpmpy
 
-# Answer Set Programming with Clingo
+# Then use with your task
+coder --with cpmpy --project coder-examples/cpmpy/cpmpy.md "Solve 8-queens"
 coder --with clingo --project coder-examples/clingo/clingo.md "Model bird flight"
 ```
 
@@ -225,7 +226,7 @@ from agentic_python_coder import get_openrouter_llm, list_available_models
 
 llm = get_openrouter_llm(model="sonnet45")
 print(list_available_models())
-# ['deepseek31', 'gemini25', 'gpt5', 'grok41', 'opus45', 'qwen3', 'sonnet45']
+# ['deepseek31', 'gemini25', 'gemini3pro', 'gpt52', 'grok41', 'opus45', 'qwen3', 'sonnet45']
 ```
 
 ---
@@ -254,17 +255,32 @@ Add to your MCP settings (e.g., `~/.claude/claude_desktop_config.json` or projec
 | Tool | Description |
 |------|-------------|
 | `python_exec` | Execute Python code. Auto-starts session if needed. Default 30s timeout. |
-| `python_reset` | Clear session state. Optionally install packages (e.g., `packages=["numpy", "pandas"]`). |
-| `python_status` | Check if session is active, Python version, installed packages, defined variables. |
+| `python_reset` | Create new kernel (no `kernel_id`) OR reset existing kernel (with `kernel_id`). Optionally install packages. |
+| `python_status` | Check session state: active flag, all active kernel IDs, Python version, packages, variables. |
 | `python_interrupt` | Send interrupt signal to stop long-running code. Session state is preserved. |
 
+### Multi-Agent Workflow
+
+For parallel agents, each agent gets its own kernel:
+
+```
+Agent A                              Agent B
+────────                             ────────
+python_reset() → kernel_id="aaa"     python_reset() → kernel_id="bbb"
+python_exec(kernel_id="aaa", ...)    python_exec(kernel_id="bbb", ...)
+python_exec(kernel_id="aaa", ...)    python_exec(kernel_id="bbb", ...)
+```
+
+Simple single-agent use: just call `python_exec()` — the default kernel auto-starts.
+
 ### Features
 
 - **Persistent state**: Variables, imports, and definitions persist across executions
-- **Auto-start**: Session starts automatically on first `python_exec`
+- **Auto-start**: Default session starts automatically on first `python_exec`
 - **Package installation**: Use `python_reset` with `packages` parameter to install dependencies
 - **Timeout handling**: Long-running code times out gracefully (session preserved)
 - **Interrupt support**: Stop runaway code without losing session state
+- **Multi-kernel**: Each `python_reset()` creates an isolated kernel for parallel agents
 
 ### Usage Tips
 

{agentic_python_coder-2.2.1 → agentic_python_coder-3.0.0}/coder/prompts/system.md

@@ -59,8 +59,13 @@ Build solutions incrementally:
 ## Important Guidelines
 
 1. **Focus on the Task**: Complete what's requested, nothing more
-2. **Test Efficiently**: One or two test cases are usually sufficient
-3. **Save Once**: Call save_code only when you have the final code
+2. **Verify Before Saving**: Before calling save_code, you MUST verify your solution:
+   - Execute the full script via python_exec and confirm it produces correct output
+   - For constraint/logic problems: write a verification function that checks the output against EVERY constraint in the problem statement using plain Python asserts, independent of your solver model
+   - For problems with a specific output format: assert that JSON keys, array shapes, and value ranges match the spec exactly
+   - For optimization: confirm optimality (e.g., re-solve with a stricter bound and confirm infeasibility)
+   - Do NOT trust that solver.solve()==True means your model is correct — your constraints may be wrong
+3. **Save Once**: Call save_code only after verification passes
 4. **Stop When Done**: Don't add features not requested
 
 ## Error Recovery
@@ -83,9 +88,10 @@ Before saving any code with save_code, your script MUST pass this checklist:
 ## Task Completion
 
 When finishing:
-1. Verify the solution works correctly
-2. Clean the code according to the **Code Cleaning Requirements** above
-3. Call save_code with the complete, cleaned code
-4. STOP - do not continue unless asked
+1. Execute the full solution and verify it produces correct, complete output
+2. For logic/constraint problems: run an independent verification that checks every constraint
+3. Clean the code according to the **Code Cleaning Requirements** above
+4. Call save_code with the complete, cleaned code
+5. STOP - do not continue unless asked
 
 Your goal is efficient, focused problem-solving.

{agentic_python_coder-2.2.1 → agentic_python_coder-3.0.0}/coder/prompts/system_todo.md

@@ -76,8 +76,13 @@ Build solutions incrementally:
 
 1. **Todo List is Mandatory**: ALWAYS use todo_write after understanding the problem
 2. **Focus on the Task**: Complete what's requested, nothing more
-3. **Test Efficiently**: One or two test cases are usually sufficient
-4. **Save Once**: Call save_code only when you have the final code
+3. **Verify Before Saving**: Before calling save_code, you MUST verify your solution:
+   - Execute the full script via python_exec and confirm it produces correct output
+   - For constraint/logic problems: write a verification function that checks the output against EVERY constraint in the problem statement using plain Python asserts, independent of your solver model
+   - For problems with a specific output format: assert that JSON keys, array shapes, and value ranges match the spec exactly
+   - For optimization: confirm optimality (e.g., re-solve with a stricter bound and confirm infeasibility)
+   - Do NOT trust that solver.solve()==True means your model is correct — your constraints may be wrong
+4. **Save Once**: Call save_code only after verification passes
 5. **Stop When Done**: Don't add features not requested
 
 ## Error Recovery
@@ -101,10 +106,11 @@ Before saving any code with save_code, your script MUST pass this checklist:
 
 When finishing (these should be your final todo items):
 1. Ensure all todo items are marked as completed
-2. Verify the solution works correctly
-3. Clean the code according to the **Code Cleaning Requirements** above
-4. Call save_code with the complete, cleaned code (final todo item)
-5. STOP - do not continue unless asked
+2. Execute the full solution and verify it produces correct, complete output
+3. For logic/constraint problems: run an independent verification that checks every constraint
+4. Clean the code according to the **Code Cleaning Requirements** above
+5. Call save_code with the complete, cleaned code (final todo item)
+6. STOP - do not continue unless asked
 
 Note: Your todo list should show a clear progression from planning through completion.
 

agentic_python_coder-3.0.0/coder/src/agentic_python_coder/__init__.py

@@ -0,0 +1,89 @@
+"""Python Coding Agent - A minimal coding assistant using direct OpenAI API and OpenRouter."""
+
+__version__ = "3.0.0"
+
+# High-level API (recommended for most users)
+from agentic_python_coder.runner import solve_task
+
+# Lower-level API (for custom workflows)
+from agentic_python_coder.agent import (
+    CodingAgent,
+    create_coding_agent,
+    run_agent,
+    get_final_response,
+    DEFAULT_STEP_LIMIT,
+)
+
+# LLM utilities
+from agentic_python_coder.llm import (
+    LLMConfig,
+    get_openrouter_llm,
+    load_model_config,
+    list_available_models,
+    DEFAULT_MODEL,
+)
+
+# Tool system
+from agentic_python_coder.tools import (
+    Tool,
+    ToolRegistry,
+    create_tool_registry,
+)
+
+# Kernel management (multi-kernel API)
+from agentic_python_coder.kernel import (
+    # Core functions
+    create_kernel,
+    execute_in_kernel,
+    shutdown_kernel_by_id,
+    interrupt_kernel_by_id,
+    restart_kernel,
+    # Query functions
+    list_kernels,
+    kernel_exists,
+    get_kernel_info,
+    shutdown_all_kernels,
+    # Backward compat
+    get_kernel,
+    shutdown_kernel,
+    # Constants
+    DEFAULT_KERNEL_ID,
+    MAX_KERNELS,
+)
+
+__all__ = [
+    # Version
+    "__version__",
+    # High-level
+    "solve_task",
+    # Low-level agent
+    "CodingAgent",
+    "create_coding_agent",
+    "run_agent",
+    "get_final_response",
+    "DEFAULT_STEP_LIMIT",
+    # LLM
+    "LLMConfig",
+    "get_openrouter_llm",
+    "load_model_config",
+    "list_available_models",
+    "DEFAULT_MODEL",
+    # Tool system
+    "Tool",
+    "ToolRegistry",
+    "create_tool_registry",
+    # Kernel management
+    "create_kernel",
+    "execute_in_kernel",
+    "shutdown_kernel_by_id",
+    "interrupt_kernel_by_id",
+    "restart_kernel",
+    "list_kernels",
+    "kernel_exists",
+    "get_kernel_info",
+    "shutdown_all_kernels",
+    "get_kernel",
+    "shutdown_kernel",
+    "DEFAULT_KERNEL_ID",
+    "MAX_KERNELS",
+]