The directory

Choose chunk sizes, overlaps, and structure-aware splits so retrieved context actually helps the model answer.

Combine BM25-style lexical search with dense embeddings and fuse results so both rare terms and paraphrases are found.

Score groundedness, relevance, and completeness with judge prompts plus deterministic checks wired into CI.

Shape tool names, parameters, descriptions, and error messages so the agent calls the right tool with valid arguments.

Bound agent autonomy with step budgets, spend caps, human gates, and checkpointing so failures stay cheap and debuggable.

Split work across specialized agents with typed handoff contracts so context survives delegation without ballooning.

Organize identity, rules, tools, and output contracts into sections that evolve without breaking behavior.

Choose and order examples covering edge cases so the model generalizes instead of copying surface patterns.

Choose and order examples covering edge cases so the model generalizes instead of copying surface patterns.

Get machine-parseable JSON reliably using schema-first prompting, constrained decoding, and repair loops.

Clean, deduplicate, and balance instruction-response pairs so fine-tuning learns behavior rather than noise.

Pick the cheapest adaptation layer that solves the problem using a decision ladder.

Allocate milliseconds across retrieval, prompting, generation, and streaming so p95 meets product targets.

Route by task complexity, cache aggressively, and enforce budgets so unit economics hold as usage grows.

Treat all retrieved content as untrusted input: isolate instructions from data, gate actions, and fuzz continuously.

Tool schema design, transport choice, error contracts, and test harness for Model Context Protocol servers.

Decide what renders on the server versus client in React Server Components to minimize JavaScript shipped.

Choose between local state, lifted state, URL state, and stores by asking who needs the data and when.

Treat non-happy paths as features: skeleton loaders, typed errors, retry affordances, useful empties.

Encode brand decisions as named tokens (CSS variables) consumed by utilities so themes change in one place.

Encode brand decisions as named tokens (CSS variables) consumed by utilities so themes change in one place.

Use fluid grids, container queries, and content-driven breakpoints instead of pixel-perfect mockups.

Animate with purpose: entrance choreography, micro-interactions, and reduced-motion respect.

Diagnose LCP, INP, and CLS from field data and apply targeted fixes per metric rather than guessing.