Personal AI
Training Flywheel
Turn your AI interaction traces into personalized training data. Import from Claude Code, ChatGPT, or any MCP tool — fine-tune a model that learns your patterns.
Your coding sessions are training data. We use them.
Every time you use Claude Code to write, debug, or refactor software, BashGym silently captures that session as a structured trace. Those traces feed an automated pipeline that fine-tunes a small model trained entirely on how you actually code — your conventions, your repos, your patterns. Over time that model takes over routine tasks from Claude, cutting your API costs and response times while getting more personalized every day.
Capture
Hooks installed into Claude Code record every tool call, file edit, and bash command as a structured execution trace — automatically, with no change to how you work.
Train
Traces are scored, sanitized, and fed into a fine-tuning pipeline. SFT, DPO, GRPO, RLVR, or distillation — you choose the strategy. The result is a small local model that knows your codebase.
Deploy
A confidence-based router progressively shifts traffic from Claude to your local model. Simple tasks go local (~50ms). Complex ones fall back to Claude. The flywheel keeps spinning.
Platform Walkthrough
See BashGym in action — from trace capture to model deployment.
11 Core Capabilities
Everything you need to capture traces, train models, and deploy your own coding assistant.
Trace Capture
Intercept Claude Code tool calls via hooks. Capture prompts, tool outputs, and reasoning traces automatically.
Quality Framework
Multi-judge scoring with syntax, semantic, and execution validators. Only high-quality traces become training data.
Privacy by Design
PII detection, secret scrubbing, and path anonymization. Your code stays private throughout the pipeline.
Training Strategies
SFT, DPO, GRPO, RLVR, and distillation pipelines. Choose the right strategy for your model and data.
Model Registry
Version, tag, and manage trained model artifacts. Track lineage from trace to deployed checkpoint.
Progressive Routing
Confidence-based routing between your local model and Claude. Your model handles what it knows, Claude handles the rest.
Real-Time Dashboard
Monitor trace collection, training progress, model performance, and routing decisions in a live dashboard.
Multi-Cloud
Train on Lambda Labs, RunPod, Vast.ai, or your own GPUs. Cloud-agnostic infrastructure provisioning.
Benchmarks
SWE-bench, HumanEval, and custom project-specific benchmarks. Measure real improvement on real tasks.
Safety Guardrails
Harmful content filtering, bias detection, and output validation. Safe models from safe data.
Orchestrator
Decompose a spec into a Task DAG, run parallel workers in isolated git worktrees, and feed results back into the training pipeline.
The Ouroboros Flywheel
A self-reinforcing loop: use Claude, capture traces, train your model, deploy it, repeat.
ACT
Use Claude Code normally
VERIFY
Judge trace quality
SYNTHESIZE
Build training data
TRAIN
Fine-tune your model
DEPLOY
Route to your model
REPEAT
Continuously improve
Live Training Monitor
Watch your model improve in real time. Track trace collection, training epochs, loss curves, and deployment status from a single dashboard.
- Trace collection stats and quality scores
- Training progress with loss and metric curves
- Model registry with version comparison
- Routing confidence and fallback rates
- Benchmark results across model versions
Three Steps to Your Own Model
Install Hooks
Install BashGym hooks into Claude Code. Traces are captured automatically as you work.
Use Claude Code Normally
Keep coding as usual. BashGym silently captures, scores, and curates high-quality training data.
Train Your Model
Launch training with one command. BashGym handles data prep, fine-tuning, evaluation, and deployment.
8-Layer Architecture
A modular system from trace capture to API serving.
Works With Your Stack
Start Training Your Own Model
Upload your traces, generate training examples, push to HuggingFace. The flywheel starts with one upload.
Get BashGym Updates
New features, training strategies, and model releases. No spam.
Frequently Asked Questions
What is BashGym?
BashGym is a self-improving agentic dev gym that captures execution traces from Claude Code sessions and converts them into fine-tuning datasets. It trains smaller, personalized language models that learn your coding patterns, conventions, and workflows through the Ouroboros Flywheel.
How does the Ouroboros Flywheel work?
The Ouroboros Flywheel is a self-reinforcing loop with six stages: Act (use Claude Code normally), Verify (judge trace quality with multi-judge scoring), Synthesize (build training data from verified traces), Train (fine-tune your model via SFT, DPO, GRPO, RLVR, or distillation), Deploy (route tasks to your model with confidence-based routing), and Repeat (continuously improve as more traces are captured).
What models can I train with BashGym?
BashGym supports fine-tuning smaller open-source language models such as CodeLlama 7B. It integrates with HuggingFace, Ollama, and NVIDIA NeMo, and supports training on Lambda Labs, RunPod, Vast.ai, or your own GPUs.
Do I need ML expertise to use BashGym?
No. BashGym is designed for developers, not ML engineers. You install hooks into Claude Code, keep coding as usual, and launch training with one command. BashGym handles data preparation, quality scoring, fine-tuning, evaluation, and deployment automatically.
How does BashGym capture training data?
BashGym installs hooks into Claude Code that silently record every tool call, file edit, and bash command as a structured execution trace. These traces are automatically scored for quality, scrubbed for PII and secrets, and curated into training datasets — with no change to how you work.
Is BashGym free?
Yes. BashGym is open source under the MIT License. You can self-host it by cloning the GitHub repository, or use the hosted web app at bashgym.fly.dev. Cloud GPU costs for training are separate and depend on your chosen provider.