All cookbook recipes

Every Stratix cookbook recipe — runnable snippets organized by the workflow stage they belong to.

Recipes are the "give me code I can paste" shape. Each one is small, self-contained, and runnable. They're organized below by the workflow stage they map to — Select, Build, Observe, Evaluate, Improve — plus cross-cutting groupings (industry, integration, migration, SDK basics) at the bottom.

If you're new, start with Top 5 starter recipes. Otherwise, browse below.

Select

Pick the model, benchmark, or prompt before you commit ECU.

• Pick a model for a task
• Compare prompts on a small test set

Build

Instrument your code and ship the wiring.

Framework adapters

• Instrument a LangChain agent
• Instrument a LlamaIndex pipeline
• Instrument a Haystack pipeline
• Instrument an OpenAI Assistants run
• Instrument a Vercel AI SDK call
• Instrument an Anthropic SDK call
• Instrument a custom Python pipeline
• OTel-style trace ingestion

CI / release

• Containerized CI runner
• Pre-release readiness eval

Observe

Capture, route, and warehouse what your system is doing.

• Backfill from existing logs
• Post-release continuous eval

Evaluate

Score what you've captured.

Score by output shape

• Score a structured-output prompt
• Score a free-form chat prompt
• Score code outputs (HumanEval-style)
• Score reasoning chains
• Score retrieval quality (precision@k)
• Score multilingual outputs
• Validate JSON output with schema

Score by failure mode

• Catch hallucinations
• Build a faithfulness judge for RAG
• Detect tool-call regressions in agents

Author your own scorers

• Custom code grader

Aggregation and gating

• Span-level scoring
• Multi-judge consensus
• Confidence-weighted aggregation
• Cost-aware evaluation
• Run a judge in CI without merging on regression

Improve

Tune what you've scored, drive cost down, keep regressions out.

• Daily smoke eval
• Quarterly model re-evaluation
• GEPA basics
• GEPA with held-out validation set
• Iteratively label and re-optimize

Cross-cutting: by industry

• Healthcare: clinical Q&A safety judge
• Financial services: advisor chatbot guardrails
• Legal: contract review faithfulness
• Retail: product Q&A scoring
• Insurance: claims triage scoring
• Education: tutor helpfulness

Cross-cutting: by CI / notification integration

• GitHub Actions gate
• GitLab CI gate
• Buildkite gate
• Jenkins gate
• Slack notifications
• PagerDuty escalation

Cross-cutting: migration from another platform

• Port a LangSmith eval
• Port a Braintrust eval
• Port an MLflow run
• Port a Promptfoo config

SDK basics

• Sync vs async clients
• Pagination over large result sets
• Retry and backoff
• Rate-limit handling

Where to next

• Top 5 starter recipes — the curated week-one subset
• All tutorials — guided, end-to-end learning paths
• All guides — task-shaped walkthroughs
• stratix-python samples — many recipes mirror runnable source samples

Integrate

• Send results to S3
• Send results to BigQuery
• Send results to Snowflake