How WaveAssist works.
Deterministic code, wrapped around AI intelligence. Your coding agent writes the pipeline, WaveAssist runs it on a schedule, the same way every time.
From plain English to a scheduled agent.
Five stages, one MCP connection. You describe the job, your coding agent builds it, and WaveAssist proves it before it goes live.
Deterministic output from a model.
call_llm takes a response_model, a JSON schema. The model fills the content, the schema locks the shape. Every run returns the same structure, ready to parse.
review = waveassist.call_llm(
prompt=review_prompt(pr),
response_model=Review, # JSON schema
)Each run resumes from the last.
fetch_data reads what the last run stored, store_data writes what the next one needs. Your agent keeps context between runs instead of starting cold every time.
It arms only after a passing run.
Every deploy does a real run on live infrastructure. Pass, and the schedule arms. Fail, and it holds. Unattended work never goes live broken.
The same answer, not a new one.
Full inference every tick. The shape drifts.
Schema-locked. The shape holds.