teich 0.1.1a9

Turn coding agent traces into training data

Teich

Turn coding agent sessions into training data.

Generate → Convert → Train

Run Codex or Pi, capture raw traces, and convert them into structured training examples for fine-tuning.

⚡ Quick Start

pip install teich

teich init my-project && cd my-project
teich generate -c config.yaml

⭐ What Teich Does

• Trace-first data collection: Run real coding agents, keep the raw session traces
• Multi-agent support: Works with Codex and Pi
• Structured output: Converts traces into chat messages with tool calls, reasoning, and tool results
• SFT-ready formatting: Applies chat templates and creates assistant masks for supervised fine-tuning
• Hugging Face integration: Load traces from local folders or dataset repos like badlogicgames/pi-mono (or any datasets generated with the tool)

📥 Install

pip install teich

Requirements for trace generation:

• Docker
• OpenAI/OpenRouter API key (or local OpenAI-compatible endpoint)

The Python utilities work without Docker if you already have traces.

🚀 Usage

Generate traces from prompts

# Initialize project
teich init my-project
cd my-project

# Add prompts to prompts.csv, then:
export OPENAI_API_KEY=sk-...
teich generate -c config.yaml

Outputs: raw traces in output/, sandboxes in sandbox/, and a README.md.

Convert traces to training data

from teich import convert_traces_to_training_data
from pathlib import Path

examples = convert_traces_to_training_data(Path("./output"))
print(examples[0]["messages"])

Load and format for training

from teich import load_traces, format_and_mask

# Load from local folder or HF dataset
dataset = load_traces("badlogicgames/pi-mono", split="train")

# Apply chat template and create masks
training_data = format_and_mask(
    dataset,
    tokenizer,
    chat_template_kwargs={"enable_thinking": True}
)

# Preview a formatted example
print(training_data.preview())

📋 Configuration

config.yaml:

agent:
  provider: codex  # or pi

model:
  model: codex-mini-latest
  approval_policy: never
  sandbox: danger-full-access

prompts_file: prompts.csv

output:
  traces_dir: ./output
  sandbox_dir: ./sandbox

Local providers (LM Studio, Ollama)

export TEICH_PROVIDER=LMstudio
export TEICH_MODEL=gemma-4
export TEICH_BASE_URL=http://localhost:1234/v1
export TEICH_API_KEY=llm

teich generate -c config.yaml

🏗️ Data Structure

Training examples include:

• prompt: initial task description
• messages: chat history (system, user, assistant, tool)
• tools: tool schemas used in the session
• metadata: session info, model, timestamps

Assistant messages capture:

• content: text response
• reasoning_content: chain-of-thought traces
• tool_calls: function calls with arguments

🔧 Python API

from teich import (
    load_traces,           # Load from folder or HF dataset
    format_and_mask,        # Apply chat template + assistant masks
    convert_traces_to_training_data,  # Convert raw traces to examples
    Config,                 # Load config.yaml
    TrainingExample         # Typed training example
)

📦 Trace-First Workflow

Teich preserves the raw agent session as the source of truth:

1. Collect: Run agents on real tasks → raw .jsonl traces
2. Inspect/Share: Traces are human-readable and uploadable
3. Convert: Transform to structured examples when ready
4. Format: Apply model-specific chat templates for training

This means you can:

• Re-convert with different logic later
• Share raw traces before releasing training data
• Train on the same sessions with different model templates

🛠️ Development

uv pip install -e ".[dev]"
pytest tests/test_formatter.py tests/test_loader.py -q

📌 Status

Teich is alpha. The core workflow is stable and usable. APIs may evolve as more agent types and training workflows are added.

📄 License

Apache-2.0