/ai--guardrails

Scaffolds a chatbot on your existing stack: streaming responses, persistent conversation history, optional RAG grounding and a feedback loop.

Implements vector embeddings in your app for semantic search, RAG or similarity matching: ingestion, chunking, querying and relevance tuning.

Builds a test harness that measures a RAG pipeline: a golden question set, retrieval metrics, groundedness checks and a regression gate.

Measures where your app spends tokens, then applies model routing, caching and prompt compression, keeping each change only if an eval shows quality holds.

Sets up an evaluation suite for an AI feature: test cases, rubrics, code-based checks, LLM-as-judge with bias controls and a CI gate.

Instruments your AI app with traces, token and cost tracking, quality signals, dashboards and alerts for every model call.

Builds a Model Context Protocol server that exposes your project's functions, APIs or data to clients like Claude Code, with transport, auth and tests.

Designs a system of several AI agents with an orchestration pattern, contracts between agents and failure handling, or advises a single agent instead.

Reviews and rewrites your system prompts, user prompts and tool definitions, then gives an A/B comparison of the original and improved versions.