CCAR-F - Agentic Architecture & Orchestration (27% of the exam) - Section 1.3
Configure subagent invocation, context passing, and spawning.
The Task tool spawns subagents, and a coordinator's allowedTools must include Task for that to work. Subagent context is explicit: nothing is inherited and no memory is shared between invocations, so prior findings travel in the prompt. Candidates should configure AgentDefinition descriptions, system prompts and tool restrictions, preserve attribution by separating content from metadata, and spawn parallel subagents by emitting multiple Task calls in a single response.
Task toolallowedTools configurationAgentDefinitionexplicit context passingparallel Task calls in one responsefork-based session management
Eight engineers on a platform team each hand write their own prompt when they dispatch a Task subagent to explain a legacy module, and each chooses the subagent's tools ad hoc. Over one month the reports arrive in six different shapes, two subagents were given Bash and spent turns installing dependencies the task did not need, and the average time a reviewer spends reading a report has risen from four minutes to eleven. The team lead wants every module explanation to arrive in one shape with one tool set. Which change most effectively addresses this?
AAsk each engineer to copy an agreed prompt template from the team wiki into the Task call, and review the pasted prompts during the weekly code review to catch any that have drifted.
BDefine a single reusable AgentDefinition for the module explainer carrying its prompt, its description and its tool set, and have every engineer invoke that definition rather than composing a prompt.check_circle Correct
CRecord the required report shape in the project CLAUDE.md so that each subagent picks the convention up from the repository, and leave the tool set to the engineer dispatching the task.
DAdd three worked examples of a well formed module explanation to each engineer's prompt, so that the expected headings and level of detail are demonstrated before the subagent begins work.
Package a repeatedly used subagent as one AgentDefinition carrying its prompt and tool set, rather than as prompts composed per caller. Both symptoms trace to the subagent being reconstructed by hand at every call site. An AgentDefinition binds the prompt, the description used for selection and the permitted tools into a single named artefact, so consistency comes from the configuration being shared rather than from every engineer reproducing it correctly.
Why A is wrong: A shared template does raise consistency and needs no engineering work, but it depends on eight people pasting the current version correctly every time, and a review held after the subagent has run cannot prevent the misconfigured tool set from being used.
Why B is correct: An AgentDefinition makes the prompt and the permitted tools one versioned artefact that every invocation resolves to, so the report shape and the tool set stop being per engineer choices and a change made once reaches all eight callers.
Why C is wrong: Repository level guidance is genuinely useful for conventions and travels with the code, but it is guidance rather than configuration and it leaves the tool set unaddressed, which is what allowed the two subagents to spend turns on dependency installation.
Why D is wrong: Examples do reliably tighten output shape when a model is unsure what is wanted, but they must still be pasted into eight separate prompts to take effect, so the drift they are meant to cure reappears through the same route it arrived by.
Exam traps in Agentic Architecture & Orchestration
Answers that look right on this material and are not. Each one is a distractor from a different question in the CCAR-F bank for this domain.
Add a routing classifier that ranks the four services by likelihood of holding the retry logic, so the most probable service is investigated first and the rest are skipped when it succeeds.
errorWhy it is wrong: Ordering the work by likelihood is a reasonable optimisation and would help in the lucky case, but it leaves the dispatch sequential, adds a classifier before simpler structural work has been tried, and returns an incomplete map whenever retry logic exists in more than one service.
The delegation prompt plus a compacted summary of the coordinator's transcript, because the Task tool condenses the parent conversation and passes it down before the subagent's first turn.
errorWhy it is wrong: Compaction is a real context management behaviour within a single session, which makes this sound familiar, but it operates on a conversation to keep it inside its own window rather than acting as a handover channel into a freshly created subagent context.
Move the coordinator to a configuration with a much larger context window, so that the delegation carries more of the conversation history through to the specialist subagent that performs the dispute analysis.
errorWhy it is wrong: Tempting because a bigger window is the usual response to an agent that appears to have lost information. It is wrong because the coordinator has lost nothing: the gap sits at the delegation boundary, and window size does not change what crosses it.