CCAR-F Agentic Architecture and Orchestration Practice Question
An architect notices that an agent using Claude 3.5 Sonnet frequently 'hallucinates' tool arguments when the tool schema is very large (over 50 parameters). What is the most effective architectural change to improve tool call accuracy?
⚠ Common exam trap
Candidates attempt to solve tool hallucination by adding more warning instructions in the system prompt, rather than addressing the root cause of bloated tool schemas.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Decompose the large tool into several smaller, specialized tools
Large, complex tool schemas can overwhelm a model's attention, leading to errors in parameter generation. By breaking down a single massive tool into several smaller, more focused tools with fewer parameters, the architect simplifies the reasoning task for the model, making it much more likely to generate valid and accurate arguments for each call.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Switch to Claude 3 Haiku to reduce the cost of the incorrect tool calls
Why it's wrong here
Switching to a smaller model like Haiku will likely decrease accuracy even further, as smaller models generally have a harder time adhering to complex JSON schemas than Sonnet or Opus. Reducing cost does not solve the underlying functional problem of the agent failing to perform its task correctly.
- ✓
Decompose the large tool into several smaller, specialized tools
Why this is correct
Breaking down a complex tool into smaller units (e.g., 'update_user_address' and 'update_user_billing' instead of 'update_entire_profile') reduces the cognitive load on the model. This modular approach makes the schemas easier to follow and results in significantly higher reliability and fewer parameter hallucinations in production agentic loops.
- ✗
Remove the 'description' fields from the JSON schema to save context space
Why it's wrong here
Descriptions are critical for the model to understand the purpose and expected format of each parameter. Removing them makes the model's job much harder and is almost certain to increase the rate of hallucinations and errors, as the model will be forced to guess the meaning of parameter names.
- ✗
Increase the 'top_p' value to 1.0 to give the model more creative freedom
Why it's wrong here
Creative freedom is the opposite of what is needed for structured tool use. Tool calling requires high precision and strict adherence to a schema. Increasing 'top_p' or temperature makes the model's output more random, which typically leads to more frequent syntax errors and invalid parameter values in the tool-use block.
About these practice questions
Courseiva writes every CCAR-F question from scratch — 271 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Anthropic exam blueprint
This CCAR-F practice question is part of Courseiva's free Anthropic certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the CCAR-F exam.