Agents And Tool Use hard
A team splits a coding agent into planner, coder, tester, and reviewer agents, and each receives the full transcript of the previous agent. The expected quality gains did not appear. Cost tripled, and the reviewer often fixates on approaches that earlier agents abandoned. When is a multi-agent split actually justified?
- A Whenever a task spans distinct roles, because specialized personas reason more accurately than a single generalist agent does.
- B When subtasks run in parallel or generate bulky context worth isolating, and each subagent returns a compact result, not a transcript.
- C When the base model is too weak for the task, since several agents critiquing one another compensate for limited capability.
- D When the transcript exceeds one context window, since handing full transcripts between agents spreads the token load across several windows.
Answer: B. When subtasks run in parallel or generate bulky context worth isolating, and each subagent returns a compact result, not a transcript.
Multi-agent splits pay off through **context isolation and parallelism**: a subagent can read 100,000 tokens of code and return a 500-token finding. Passing full transcripts throws that benefit away, multiplying cost and importing irrelevant history. The cost is coordination overhead and lost shared context, so a single agent is usually better for tightly sequential work like this. Role personas do not make the model smarter.
One system design question a day, with spaced repetition so the ones you miss come back.
Get the appRelated questions
- You are defining a 'schedule_meeting' tool for an agent. Which schema decision most reduces wrong or malformed calls?
- A model emits three parallel tool calls: `get_inventory`, `get_price`, and `get_shipping`. To cut latency, the host appends each result to the transcript as it arrives and immediately re-invokes the model. Runs start failing with malformed-request errors, and some produce hallucinated shipping costs. What should the host do?
- A support agent should remember each customer's preferences across months of sessions. The team appends every past session summary to the system prompt. After six months, prompts exceed 80,000 tokens, latency has doubled, and the agent sometimes ships to an address the customer replaced weeks ago. What redesign fixes both problems?
- A teammate proposes splitting a single coding agent into five agents, arguing that more agents means smarter results. When is a multi-agent split actually justified?
- A data-analysis agent runs model-generated Python inside the same container as your API service. That container has database admin credentials in environment variables. A security review flags this as critical. What is the best mitigation?
- An internal agent has grown to 120 tools, many with overlapping names like search_docs, find_docs, and query_docs. Wrong-tool errors are rising. What is the best next step?