Cheatsheet
This cheatsheet gives copy-ready commands and code snippets for common workflows.
Use it during local setup, debugging, and interview prep drills.
Local development commands
| Task | Command |
|---|---|
| Install docs dependencies | pip install -r requirements.txt |
| Start docs server | mkdocs serve |
| Build docs | mkdocs build |
| Run API server | uvicorn src.api.server:app --reload |
API smoke test
| Goal | Command |
|---|---|
Test /chat endpoint |
curl -X POST http://localhost:8000/chat -H "Content-Type: application/json" -d '{"prompt":"add 2 and 3"}' |
Python snippets
from src.memory.store import MemoryStore
store = MemoryStore()
store.save("hello", "world")
print(store.all())
Troubleshooting matrix
| Symptom | Likely cause | First action |
|---|---|---|
| "No suitable tool found" | Router rule too narrow | Expand intent checks in route |
| Wrong arithmetic output | Incorrect argument extraction | Log args before tool execution |
| Missing trace visibility | Trace sink too minimal | Replace print with structured logger |
Should I optimize prompts or architecture first?
Optimize architecture first. Clear contracts and telemetry make prompt tuning far more effective.
What is the best first automation task?
Add a repeatable smoke test that calls /chat with representative prompts.
It catches integration regressions early.
--8<-- "_abbreviations.md"