# Playbooks — recommendations for AI projects (wpnews)

Source-cited practices grouped by problem. Each playbook has a Markdown and JSON twin.

- [Working with AI coding agents](https://wpnews.pro/playbooks/coding-agents.md) — 4 recommendations. Getting reliable work out of coding agents: instruction and context files, task scoping, reviewing agent-written code, repository conventions, hooks and workflow guardrails.
- [Designing agent systems](https://wpnews.pro/playbooks/agent-architecture.md) — 2 recommendations. Structuring agents: orchestration and multi-agent patterns, tool and MCP server design, memory, planning loops, state and handoffs.
- [Securing agents and AI tools](https://wpnews.pro/playbooks/agent-security.md) — 2 recommendations. Sandboxing, scoped credentials, prompt-injection defence, permission and approval gates, egress control, supply chain of models and tools.
- [Verifying model output](https://wpnews.pro/playbooks/output-verification.md) — 0 recommendations. Grounding and checking what a model claims: citation and quote verification, hallucination detection, LLM-as-judge pitfalls, structured-output validation, human approval of actions.
- [Evals and testing for LLM apps](https://wpnews.pro/playbooks/evals-testing.md) — 1 recommendations. Measuring and regression-testing LLM apps and agents: eval datasets, benchmarks you run yourself, CI gates, tests on tool calls and traces.
- [Retrieval and RAG](https://wpnews.pro/playbooks/rag-retrieval.md) — 0 recommendations. Chunking, embeddings, keyword and hybrid search, reranking, vector stores, filtering, and measuring retrieval quality.
- [Controlling LLM cost](https://wpnews.pro/playbooks/llm-cost.md) — 4 recommendations. Cutting token spend and API bills: model routing, prompt and KV caching, batching, context budgets, moving work from models to code.
- [Inference and self-hosting](https://wpnews.pro/playbooks/inference-serving.md) — 1 recommendations. Running models yourself: GPU memory, quantization, KV cache, serving engines, throughput and latency tuning, hardware choices.
- [Running AI in production](https://wpnews.pro/playbooks/production-ops.md) — 2 recommendations. Monitoring, tracing and alerting for LLM apps, incident response, rollouts, model upgrades and provider migrations, reliability.
- [Fine-tuning and data](https://wpnews.pro/playbooks/training-data.md) — 0 recommendations. Fine-tuning, distillation, RL and preference tuning, synthetic data, dataset curation and labeling for models you train.
