CCAR-P Advanced Agentic Architecture Practice Question
An agentic system uses a supervisor agent that delegates to specialized worker agents. During a long incident, the supervisor's own context fills with worker transcripts, and it begins losing track of which workers have completed and which are still running. Which TWO architectural changes best preserve the supervisor's ability to coordinate correctly? (Choose two.)
⚠ Common exam trap
The trap here is believing that a bigger context window or a restart can substitute for externalizing task state, when both leave the supervisor's coordination record coupled to transcript volume.
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
✓
Have the supervisor maintain an external task ledger that records each delegated task, its assigned worker, and its status, and read from that ledger instead of retaining full worker transcripts.
The supervisor's difficulty comes from mixing coordination state with raw worker output. Externalizing task state into a ledger and having workers return compact structured envelopes keeps the authoritative record of what is delegated and what has finished separate from bulky transcripts. The supervisor can then coordinate from small, reliable signals and pull detail only when needed, rather than trying to hold the entire incident in its context window.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Restart the supervisor with a fresh context whenever its window fills, and rely on workers to resend their current status on request.
Why it's wrong here
Restarting discards the supervisor's accumulated understanding of the incident and any in-flight decisions, causing duplicated or conflicting delegations. Relying on workers to resend status assumes they are reachable and consistent, which is fragile during an incident. This is a disruptive recovery action rather than a design that keeps coordination correct as context grows.
- ✓
Have the supervisor maintain an external task ledger that records each delegated task, its assigned worker, and its status, and read from that ledger instead of retaining full worker transcripts.
Why this is correct
An external ledger keeps authoritative state about what was delegated and what finished, so the supervisor can coordinate without carrying every worker transcript in context. It survives context growth and lets the supervisor query only the fields it needs. This directly addresses losing track of completed versus running workers while keeping the supervisor's context bounded.
- ✗
Allow workers to message each other directly to resolve dependencies, removing the supervisor from the coordination path entirely.
Why it's wrong here
Peer-to-peer messaging removes the supervisor's visibility into task state, which is the exact problem being solved. Dependencies may resolve faster in some cases, but the supervisor can no longer tell which workers are done or stuck, and debugging distributed failures becomes harder. This trades a coordination bottleneck for a coordination blind spot.
- ✗
Increase the supervisor's context window to the largest available model so it can retain all worker transcripts for the duration of the incident.
Why it's wrong here
A larger window delays the overflow but does not bound it; a long incident can generate transcripts exceeding any fixed window. Retaining full transcripts also wastes attention on irrelevant detail, making status tracking harder rather than easier. This treats the symptom by buying headroom instead of separating coordination state from raw transcript volume.
- ✓
Have each worker return a compact structured status envelope containing task id, state, and a short result summary, and have the supervisor track tasks by these envelopes.
Why this is correct
Structured status envelopes give the supervisor exactly the fields needed to know which tasks are done, running, or failed, without the bulk of full transcripts. Summaries keep context small while preserving the ability to escalate to detail on demand. This makes coordination state explicit and reliable even as the incident lengthens.
About these practice questions
Courseiva writes every CCAR-P question from scratch — 262 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-P 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-P exam.