continualcode 0.1.0__tar.gz

continualcode-0.1.0/.gitignore

@@ -0,0 +1,11 @@
+__pycache__/
+*.py[cod]
+
+.DS_Store
+
+.env
+.venv/
+venv/
+
+training_data/*.jsonl
+preference_data/*.jsonl

continualcode-0.1.0/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 Surya Dantuluri
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

continualcode-0.1.0/PKG-INFO

@@ -0,0 +1,115 @@
+Metadata-Version: 2.4
+Name: continualcode
+Version: 0.1.0
+Summary: Human-in-the-loop coding agent with online learning via prompt distillation
+Project-URL: Homepage, https://github.com/sdan/continualcode
+Project-URL: Repository, https://github.com/sdan/continualcode
+Author-email: Surya Dantuluri <surya@prava.dev>
+License-Expression: MIT
+License-File: LICENSE
+Keywords: claude-code,coding-agent,context-distillation,fine-tuning,llm,prompt-distillation,rlhf,tui
+Classifier: Development Status :: 3 - Alpha
+Classifier: Environment :: Console
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
+Classifier: Topic :: Software Development
+Requires-Python: >=3.10
+Requires-Dist: textual>=0.40.0
+Requires-Dist: tinker-cookbook>=0.1.0
+Requires-Dist: torch>=2.0.0
+Provides-Extra: dev
+Requires-Dist: pytest>=7.0.0; extra == 'dev'
+Requires-Dist: ruff>=0.1.0; extra == 'dev'
+Description-Content-Type: text/markdown
+
+# ContinualCode
+
+Human-in-the-loop coding agent with online learning via prompt distillation.
+
+A Claude Code-style TUI that learns from your approvals. When you approve or edit a tool call, the model trains on that feedback in real-time.
+
+## Features
+
+- **Claude Code-style approval UI** - 4-option selection with keyboard navigation
+- **Online prompt distillation** - Model learns to internalize policy without needing the policy prompt
+- **Permission memory** - "Don't ask again" for trusted directories
+- **Correction feedback** - Type what the model should have done (for DPO training)
+- **Checkpoint save/load** - Ctrl+S to save, resume later with `--checkpoint`
+
+## Install
+
+```bash
+pip install continualcode
+```
+
+## Quick Start
+
+```bash
+# Run the TUI
+continualcode
+
+# With a custom policy file
+continualcode --policy ./my_rules.md
+
+# Resume from checkpoint
+continualcode --checkpoint ./checkpoints/step_000100
+
+# Inference only (no training)
+continualcode --no-training
+```
+
+## Environment Variables
+
+| Variable | Default | Description |
+|----------|---------|-------------|
+| `MODEL` | `Qwen/Qwen3-4B-Instruct-2507` | Base model |
+| `TINKER_URL` | `None` | Tinker service URL |
+| `TINKER_API_KEY` | `None` | Tinker API key |
+| `ENABLE_TRAINING` | `1` | Enable LoRA training |
+| `LEARNING_RATE` | `1e-5` | Adam learning rate |
+| `LORA_RANK` | `32` | LoRA rank |
+| `POLICY_PATH` | `./policy_memory.md` | Policy file injected into teacher prefix |
+| `DISTILL_MODE` | `on_policy` | `on_policy` or `off_policy` |
+| `CHECKPOINT_DIR` | `./checkpoints` | Where Ctrl+S saves checkpoints |
+| `LOAD_CHECKPOINT` | `""` | Checkpoint to load on startup |
+
+## How It Works
+
+**Prompt distillation** trains a model to behave *as if* it had access to a long policy prompt, without actually needing it at inference time:
+
+1. **Generation** uses a "teacher" prefix that includes your policy file
+2. **Training** uses a "student" prefix without the policy
+3. The model learns to produce the same (approved) outputs without the policy context
+
+Two distillation modes:
+- `off_policy`: Classic SFT on approved outputs
+- `on_policy`: Sample from student, score with teacher logprobs, update via importance sampling
+
+## Keybindings
+
+| Key | Action |
+|-----|--------|
+| `Enter` | Send message |
+| `Ctrl+S` | Save checkpoint |
+| `Ctrl+L` | Clear conversation |
+| `Ctrl+C` | Quit |
+| `?` | Show help |
+
+**Approval UI:**
+| Key | Action |
+|-----|--------|
+| `1` / `y` | Approve |
+| `2` / `Shift+Tab` | Approve + remember |
+| `3` / `n` / `Esc` | Deny |
+| `4` / `Tab` | Type correction |
+| `e` | Edit args in $EDITOR |
+| Arrow keys | Navigate options |
+
+## License
+
+MIT

continualcode-0.1.0/README.md

@@ -0,0 +1,86 @@
+# ContinualCode
+
+Human-in-the-loop coding agent with online learning via prompt distillation.
+
+A Claude Code-style TUI that learns from your approvals. When you approve or edit a tool call, the model trains on that feedback in real-time.
+
+## Features
+
+- **Claude Code-style approval UI** - 4-option selection with keyboard navigation
+- **Online prompt distillation** - Model learns to internalize policy without needing the policy prompt
+- **Permission memory** - "Don't ask again" for trusted directories
+- **Correction feedback** - Type what the model should have done (for DPO training)
+- **Checkpoint save/load** - Ctrl+S to save, resume later with `--checkpoint`
+
+## Install
+
+```bash
+pip install continualcode
+```
+
+## Quick Start
+
+```bash
+# Run the TUI
+continualcode
+
+# With a custom policy file
+continualcode --policy ./my_rules.md
+
+# Resume from checkpoint
+continualcode --checkpoint ./checkpoints/step_000100
+
+# Inference only (no training)
+continualcode --no-training
+```
+
+## Environment Variables
+
+| Variable | Default | Description |
+|----------|---------|-------------|
+| `MODEL` | `Qwen/Qwen3-4B-Instruct-2507` | Base model |
+| `TINKER_URL` | `None` | Tinker service URL |
+| `TINKER_API_KEY` | `None` | Tinker API key |
+| `ENABLE_TRAINING` | `1` | Enable LoRA training |
+| `LEARNING_RATE` | `1e-5` | Adam learning rate |
+| `LORA_RANK` | `32` | LoRA rank |
+| `POLICY_PATH` | `./policy_memory.md` | Policy file injected into teacher prefix |
+| `DISTILL_MODE` | `on_policy` | `on_policy` or `off_policy` |
+| `CHECKPOINT_DIR` | `./checkpoints` | Where Ctrl+S saves checkpoints |
+| `LOAD_CHECKPOINT` | `""` | Checkpoint to load on startup |
+
+## How It Works
+
+**Prompt distillation** trains a model to behave *as if* it had access to a long policy prompt, without actually needing it at inference time:
+
+1. **Generation** uses a "teacher" prefix that includes your policy file
+2. **Training** uses a "student" prefix without the policy
+3. The model learns to produce the same (approved) outputs without the policy context
+
+Two distillation modes:
+- `off_policy`: Classic SFT on approved outputs
+- `on_policy`: Sample from student, score with teacher logprobs, update via importance sampling
+
+## Keybindings
+
+| Key | Action |
+|-----|--------|
+| `Enter` | Send message |
+| `Ctrl+S` | Save checkpoint |
+| `Ctrl+L` | Clear conversation |
+| `Ctrl+C` | Quit |
+| `?` | Show help |
+
+**Approval UI:**
+| Key | Action |
+|-----|--------|
+| `1` / `y` | Approve |
+| `2` / `Shift+Tab` | Approve + remember |
+| `3` / `n` / `Esc` | Deny |
+| `4` / `Tab` | Type correction |
+| `e` | Edit args in $EDITOR |
+| Arrow keys | Navigate options |
+
+## License
+
+MIT

continualcode-0.1.0/demo_project/README.md

@@ -0,0 +1,47 @@
+# Demo: Async Worker Pool Bug
+
+A subtle concurrency bug that would trip up experienced systems programmers.
+
+## The Bug
+
+`async_pool.py` has a race condition in task submission. The bug:
+
+```python
+async def submit(self, coro_fn, *args, **kwargs) -> int:
+    task_id = self._task_counter
+    self._task_counter += 1
+
+    asyncio.create_task(self._run_task(task_id, coro_fn, *args, **kwargs))
+    return task_id
+```
+
+**Problem**: `create_task()` schedules the coroutine but doesn't `await` it. The task
+starts running immediately (or on next tick). If the task completes *before* the caller
+adds `task_id` to their tracking list, and `gather_all` is called right after, there's
+a window where results can be missed.
+
+More subtly: the `_task_counter` increment and task creation aren't atomic with respect
+to the event loop. Under high concurrency with very fast tasks, IDs can get reused or
+results can land in `_results` before the caller even knows what ID they got.
+
+## How to Trigger
+
+Run with many fast tasks:
+```bash
+python async_pool.py
+```
+
+The bug is intermittent - it depends on task timing and event loop scheduling.
+
+## The Fix
+
+The `submit()` method should await the task being tracked properly, or use a different
+pattern like returning a `Future` that the caller can await, rather than an ID-based
+lookup system that races with task completion.
+
+## Demo Prompt
+
+```
+There's a race condition in async_pool.py where tasks occasionally get lost under
+high concurrency. The pool reports fewer results than tasks submitted. Find and fix it.
+```

continualcode-0.1.0/pyproject.toml

@@ -0,0 +1,71 @@
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[project]
+name = "continualcode"
+version = "0.1.0"
+description = "Human-in-the-loop coding agent with online learning via prompt distillation"
+readme = "README.md"
+license = "MIT"
+authors = [
+    { name = "Surya Dantuluri", email = "surya@prava.dev" }
+]
+keywords = [
+    "llm",
+    "coding-agent",
+    "fine-tuning",
+    "rlhf",
+    "prompt-distillation",
+    "context-distillation",
+    "tui",
+    "claude-code",
+]
+classifiers = [
+    "Development Status :: 3 - Alpha",
+    "Environment :: Console",
+    "Intended Audience :: Developers",
+    "License :: OSI Approved :: MIT License",
+    "Programming Language :: Python :: 3",
+    "Programming Language :: Python :: 3.10",
+    "Programming Language :: Python :: 3.11",
+    "Programming Language :: Python :: 3.12",
+    "Topic :: Software Development",
+    "Topic :: Scientific/Engineering :: Artificial Intelligence",
+]
+requires-python = ">=3.10"
+dependencies = [
+    "textual>=0.40.0",
+    "tinker-cookbook>=0.1.0",
+    "torch>=2.0.0",
+]
+
+[project.optional-dependencies]
+dev = [
+    "pytest>=7.0.0",
+    "ruff>=0.1.0",
+]
+
+[project.urls]
+Homepage = "https://github.com/sdan/continualcode"
+Repository = "https://github.com/sdan/continualcode"
+
+[project.scripts]
+continualcode = "continualcode.cli:main"
+
+[tool.hatch.build.targets.wheel]
+packages = ["src/continualcode"]
+
+[tool.hatch.build.targets.sdist]
+include = [
+    "src/continualcode",
+    "README.md",
+    "LICENSE",
+]
+
+[tool.ruff]
+line-length = 100
+target-version = "py310"
+
+[tool.ruff.lint]
+select = ["E", "F", "I", "W"]

continualcode-0.1.0/src/continualcode/__init__.py

@@ -0,0 +1,17 @@
+"""
+ContinualCode - Human-in-the-loop coding agent with online learning.
+
+A Claude Code-style TUI that learns from your approvals via prompt distillation.
+"""
+
+__version__ = "0.1.0"
+
+from continualcode.session import ContextDistillSession
+from continualcode.tools import TOOL_SPECS, READONLY_TOOLS, execute_tool
+
+__all__ = [
+    "ContextDistillSession",
+    "TOOL_SPECS",
+    "READONLY_TOOLS",
+    "execute_tool",
+]

continualcode-0.1.0/src/continualcode/cli.py

@@ -0,0 +1,67 @@
+#!/usr/bin/env python3
+"""
+CLI entry point for ContinualCode.
+
+Usage:
+    continualcode                    # Run the TUI
+    continualcode --help             # Show help
+    continualcode --version          # Show version
+"""
+
+import argparse
+import sys
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(
+        prog="continualcode",
+        description="Human-in-the-loop coding agent with online learning via prompt distillation",
+    )
+    parser.add_argument(
+        "--version",
+        action="version",
+        version="%(prog)s 0.1.0",
+    )
+    parser.add_argument(
+        "--model",
+        default=None,
+        help="Override MODEL environment variable",
+    )
+    parser.add_argument(
+        "--policy",
+        default=None,
+        help="Path to policy file (default: ./policy_memory.md)",
+    )
+    parser.add_argument(
+        "--checkpoint",
+        default=None,
+        help="Load checkpoint on startup",
+    )
+    parser.add_argument(
+        "--no-training",
+        action="store_true",
+        help="Disable training (inference only)",
+    )
+
+    args = parser.parse_args()
+
+    # Set environment variables from CLI args
+    import os
+    if args.model:
+        os.environ["MODEL"] = args.model
+    if args.policy:
+        os.environ["POLICY_PATH"] = args.policy
+    if args.checkpoint:
+        os.environ["LOAD_CHECKPOINT"] = args.checkpoint
+    if args.no_training:
+        os.environ["ENABLE_TRAINING"] = "0"
+
+    # Import and run the TUI
+    from continualcode.tui import TinkerCodeApp
+    app = TinkerCodeApp()
+    app.run()
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())