ShellPilot is an AI-powered command-line assistant that diagnoses terminal errors, suggests fixes, and learns from previous failures to resolve recurring issues instantly.
Run commands with:
ai-run <command>
Example:
ai-run python3 -m pip install -e
ShellPilot stores previously resolved errors and retrieves them instantly the next time they occur.
Uses Amazon Bedrock LLMs to analyze terminal failures and generate safe fixes.
Uses Titan text embeddings to match similar errors even if they aren't identical.
Once a fix is learned:
⚡ Instant fix from ShellPilot memory
ShellPilot never executes commands automatically. Users confirm before execution.
ai-run stats
Example:
ShellPilot Stats
----------------
Errors learned: 7
Top learned fixes:
• pip install -e .
• npm install --legacy-peer-deps
• sudo docker run
Terminal Error
│
▼
ShellPilot CLI
│
▼
Context Builder
│
▼
API Gateway
│
▼
AWS Lambda
│
├── Titan Embeddings (semantic search)
├── Llama 3 (fix generation)
└── DynamoDB (persistent memory)
Component Technology
CLI Tool Python AI Models Amazon Bedrock (Llama 3) Embeddings Titan Text Embeddings Backend AWS Lambda API API Gateway Memory Store DynamoDB Similarity Search Cosine Similarity
Clone the repository:
git clone https://github.com/Jaydeng75/shellpilot
cd shellpilot
Install dependencies:
pip install -e .
Run commands with AI debugging:
ai-run <command>
Example:
ai-run python3 -m pip install -e
- Run a failing command
- ShellPilot diagnoses the issue
- Suggested fix appears
- Apply fix
- ShellPilot learns it for future runs
Developers repeatedly encounter the same terminal errors. ShellPilot eliminates repetitive debugging by creating a persistent knowledge base of fixes.
Instead of repeatedly diagnosing the same problems, ShellPilot:
- Understands the error
- Suggests a fix
- Learns the solution
- Resolves the issue instantly next time
- Vector database for large-scale error memory
- Automated fix verification in sandbox environments
- IDE integration
- Multi-agent debugging architecture
- Shared team debugging knowledge bases
Built as a hackathon project demonstrating how AI + semantic search + persistent memory can transform developer debugging workflows.
ShellPilot transforms terminal debugging from:
search → guess → retry
into:
AI diagnose → apply fix → learn forever