← Lab Overview Claude Foundations

Quick Reference Sheet

Essential terms, limits, ZDR rules, and cost modifiers for the exam. Keep this open while you review.

1. Zero Data Retention (ZDR) & Compliance

ZDR ensures data is not stored after an API response is returned. Eligibility is determined per feature.

ZDR-Eligible

  • Adaptive Thinking / Effort
  • Citations
  • Compaction
  • Standard Web Search / Web Fetch
  • Structured Outputs (prompts/results only)

NOT ZDR-Eligible

  • Message Batches (stored 29 days)
  • Files API
  • Code Execution
  • MCP Connector (server-side routing)

HIPAA schema safety: JSON schemas are cached for 24 hours separately from message content. PHI must never appear in tool names, property names, or descriptions; use generic fields like record_id instead of patient_ssn.

2. Core Technical & API Limits

Feature Limit / Value
Context Windows1M tokens (Opus 4.7/4.6, Sonnet 4.6) • 200k tokens (most other active models)
Batch Max Tokens300,000 using the output-300k-2026-03-24 header
Strict Tool LimitsMax 20 strict tools • 24 optional parameters • 16 union types (for example anyOf) per request
Description LengthTool and Skill descriptions must be under 1024 characters
Message BatchesMax 100,000 requests or 256 MB per batch • Results expire after 29 days
Compaction TriggerMinimum 50,000 tokens

3. Agentic Loop & Orchestration Signals

The orchestrator must drive control flow using structured API signals rather than natural language parsing.

Signal / PatternMeaning
stop_reason == "tool_use"The loop continues. Execute the tool, append the tool_result to history, and re-call the API.
stop_reason == "end_turn"The only reliable signal for termination.
Task ToolMechanism for spawning subagents. Coordinator allowed_tools must include "Task" to delegate.
Findings → Prompt ContractSubagents are isolated by default and do not inherit the coordinator's history. Coordinators must explicitly pass findings into the subagent's prompt.

4. Structured Error & Recovery Framework

Every tool should return structured metadata to enable intelligent coordinator-level recovery.

CategoryRecovery Behavior
TRANSIENTTimeouts or service outages; Claude should attempt a retry.
VALIDATIONInvalid input, such as bad email format; agent should clarify with the user.
PERMISSIONAuthorization gaps; agent should inform the user.
BUSINESSPolicy violations, such as refund over limit; agent should communicate the rule.
isRetryableBoolean metadata that prevents the model from wasting turns on non-fixable errors.

5. Prompting & Schema Design Patterns

PatternUse
Tool Description Pattern[Input Shape] + [Canonical Use Case] + [Boundary / DO NOT use for X]. Eliminates misrouting between similar tools.
Reasoning-Led Few-ShottingInclude a Why: line in examples to teach the underlying logic so the model generalizes to novel cases.
Nullable FieldsUse ["string", "null"] to allow null when data is missing from a source, preventing hallucinations.
Extensible EnumsUse an enum field with an other option paired with a separate detail string field.

6. Repository & Workflow Governance

Governance AreaRule
CLAUDE.md HierarchyUser (~/.claude/CLAUDE.md) → Project (root/CLAUDE.md) → Directory (subdir/CLAUDE.md). Directory-level files override project-level files.
Path-Scoped RulesLocated in .claude/rules/ using glob patterns such as **/*.test.tsx to load conventions only when matching files are edited.
Progressive DisclosureThree-level skill structure: YAML frontmatter for trigger conditions, SKILL.md body for actionable steps, linked files for deep reference docs.
Session StateResume if context is valid; fork to compare divergent what-if approaches; fresh start if context is stale or misleading.

7. Reliability & Evaluation Checklist

Checklist ItemWhy It Matters
Case Facts PatternPlace non-negotiable transactional data, such as Max Budget, in a persistent block outside summarized history so it survives compaction.
Stratified Random SamplingMeasure error rates across document or prospect segments, such as small startups versus large firms, so aggregate accuracy does not mask a weak segment.
Independent Review InstanceFinal outputs should be audited by a fresh Claude session isolated from the generator's reasoning bias.
Information ProvenanceRequire subagents to output structured claim-source mappings, including URLs, excerpts, and dates, for downstream synthesis.
Back to Module 11 Lab Overview