CSCI 576 Multi-Agent Researcher — Group 6

A multi-agent literature research assistant built with Google ADK. Given a research topic, it searches ArXiv, presents papers for approval, runs parallel paper analysis, validates each summary, and produces a final synthesized literature review.

---

Setup

python -m venv .venv
source .venv/bin/activate        # Windows: .\.venv\Scripts\activate
pip install -r requirements.txt
cp .env.example .env             # then fill in your keys

Required .env keys

Variable	Required	Notes
GOOGLE_API_KEY	Yes	Gemini API access
SEED_PAPER_COUNT	No	Number of papers to research (default: 3)
SEMANTIC_SCHOLAR_API_KEY	No	Reduces throttling on paper lookups

---

Running

TUI Dashboard (recommended)

cd multi_agent_researcher
python dashboard.py

Launches a Textual TUI with:

• Left panel — live per-agent status (Planner, Researchers, Validators, Synthesizer)
• Right panel — scrollable activity log with full tool-level output
• Input bar — multi-turn conversation with the agent pipeline
• Ctrl+L to clear the log, Ctrl+C to quit

Plain CLI (original)

cd multi_agent_researcher
adk run .

---

Workflow

1. Enter a research topic — Root refines it and asks for a green light
2. Green light — Planner searches ArXiv and presents papers
3. Approve papers — type approved or remove papers by number
4. Pipeline runs automatically — researchers analyze each paper in parallel, validators check each summary, synthesizer produces the final report
5. Output path shown — synthesis report location printed on completion

Continuing a previous run

Continue from the latest planner run

Root loads the existing manifest and skips straight to the research pipeline.

---

Output structure

outputs/
  run_<timestamp>/
    planner_manifest.json
    papers/
      <arxiv_id>.pdf
      file_cache.json
    researchers/
      researcher_1/
        summary.md
        validator/
          validation_criteria.json
          validation_summary.md
      researcher_2/ ...
      researcher_3/ ...
    synthesis/
      synthesis_report.md
      synthesis_summary.json

Only the 3 most recent runs are kept (configurable via keep_last in create_run_output_dir).

---

Agent hierarchy

Root
└── Planner
    └── Research Pipeline (SequentialAgent)
        ├── Researcher Pool (parallel LoopAgents)
        │   ├── Researcher_1 + Validator_1
        │   ├── Researcher_2 + Validator_2
        │   └── Researcher_N + Validator_N
        └── Synthesizer

All tools are defined in tools/agent_tools.py. Agent prompts live in <subagent>/agent_prompt.md.