Can we use mix of Sequential and Parallel agents? Or custom agents are more suitable

[hasit73]

I’ve built a multi-agent system composed of the following agents:

file_read_agent – Reads my resume from the local system.

file_formatter_agent – Converts the text-based resume into a JSON format.

resume_parser_agent (sequential) – Calls file_read_agent and file_formatter_agent in sequence to produce a structured JSON version of my resume.

job_posting_retrieval – Retrieves the latest job postings from platforms like Naukri, LinkedIn, and Indeed using the jobspy module (no traditional web search involved).

parallel_agent – Calls both resume_parser_agent and job_posting_retrieval in parallel to gather resume and job data concurrently.

job_match_scorer_agent – Compares each job posting with my resume and assigns a match score.

presenter_agent – Formats and presents the final output in a structured manner.

root_agent – Orchestrates the overall process by calling parallel_agent, job_match_scorer_agent, and presenter_agent sequentially.

When I ask a query like:
"Can you give me 10 recently posted job postings related to Python and JavaScript?"
— the system often responds with something like "I’m not capable of doing web search," and only selectively calls one or two agents rather than executing the full chain as defined.

I’m trying to determine the root cause of this issue. Is it due to incomplete or unclear agent descriptions/instructions? Or do I need a dedicated coordinator agent that interprets user queries and ensures all relevant agents are executed in the proper sequence and context?