Claude API (platform.claude.com) — Official reference covering prompt engineering techniques and model-specific prompting differences, evaluation design, grading, guardrails, and implementation guides for typical use cases.
For details on API calls themselves, such as the Messages API, structured outputs, and streaming, see anthropic-api-core area.
Directory structure
skills/anthropic-prompt-eval/
SKILL.md
references/
prompt-engineering/
README.md
overview.md
claude-prompting-best-practices.md
prompting-claude-fable-5.md
prompting-claude-opus-5.md
prompting-claude-opus-4-8.md
prompting-claude-sonnet-5.md
test-evaluate/
README.md
develop-tests.md
handle-streaming-refusals.md
increase-consistency.md
mitigate-jailbreaks.md
reduce-hallucinations.md
reduce-latency.md
reduce-prompt-leak.md
use-cases/
README.md
overview.md
classification.md
content-moderation.md
customer-support-chat.md
legal-summarization.md
system-prompts.md
ticket-routing.md
samples/
README.md
prompt-template-structure.md
classification-prompt.md
rag-summarization-prompt.md
guardrails-prompts.md
eval-rubric-workflow.md
model-specific-prompting.md
Exploration procedure
Derive the category from the task, and identify the target page in the category's README.md:
- Find the category corresponding to the task in the mapping table below
- of that category
references/{category}/README.md（samples/(the README.md directly under it) to identify the target page - Read the target page's
.mdto confirm the details
Task → Category Mapping
| Task | Category | Reference README |
|---|---|---|
| I want to know the basic techniques of prompt design (clarity, examples, XML tags, thinking, agentic systems) | prompt-engineering | references/prompt-engineering/README.md |
| I want to know the model-specific prompting differences and migration methods for Fable 5 / Opus 5 / Opus 4.8 / Sonnet 5 | prompt-engineering | references/prompt-engineering/README.md |
| I want to know how to define success criteria, build evaluations (evals), and use code-graded / LLM-graded appropriately | test-evaluate | references/test-evaluate/README.md |
| I want to know how to detect streaming refusals, improve output consistency, and counter jailbreak / prompt injection / hallucination / prompt leak / latency | test-evaluate | references/test-evaluate/README.md |
| I want to know implementation guides for classification, content moderation, customer support chat, legal summarization, and ticket routing | use-cases | references/use-cases/README.md |
| I want to know the release notes for system prompts of the claude.ai / iOS / Android apps | use-cases | references/use-cases/README.md |
| I want to know typical usage (prompt template structure, classification + eval, summarization, guardrails, eval rubric, model-specific tuning examples) | samples | samples/README.md |