Thesis Advisor

Thesis Advisor — Thesis Writing Support Full Pipeline

An agent team collaborates to perform topic selection, literature review, methodology design, writing, and proofreading.

Execution Mode

Agent Team — 5 agents communicate directly via SendMessage and cross-verify each other's work.

Agent Roster

Workflow

Phase 1: Preparation (Orchestrator performs directly)

1. Extract from user input:
   • Discipline: Major and sub-field
   • Thesis type: Master's / Doctoral / Journal article / Undergraduate thesis
   • Topic of interest (optional): Specific topic or keywords
   • Advisor's research area (optional): Lab research direction
   • Existing materials (optional): Existing manuscripts, reference lists, data
   • Deadline (optional): Submission date
2. Create a _workspace/ directory at the project root
3. Organize the input and save it to _workspace/00_input.md
4. If existing materials are provided, copy them to _workspace/ and adjust the relevant phase
5. Determine the execution mode based on the scope of the request

Phase 2: Team Assembly and Execution

Inter-agent communication flow:

• topic-explorer completes -> sends research questions and keywords to literature-analyst; sends hypotheses and variables to methodology-expert
• literature-analyst completes -> sends methodology trends from prior studies to methodology-expert; sends literature review content to writing-coach
• methodology-expert completes -> sends full methodology design to writing-coach
• writing-coach completes -> sends completed draft to proofreader
• proofreader completes -> on critical findings, requests corrections from the relevant agent (max 2 rounds)

Phase 3: Integration and Final Report

1. Verify all files in _workspace/
2. Confirm that all critical proofreading items have been addressed
3. Report the final summary to the user

Task-Scale Modes

Existing material usage: If the user provides existing manuscripts, reference lists, etc., skip or augment the relevant phase.

Data Transfer Protocol

File naming convention: {order}_{agent}_{deliverable}.{extension}

Error Handling

Test Scenarios

Normal Flow

Prompt: "I want to write an education master's thesis. I'm interested in the impact of flipped learning on learning motivation." Expected Results:

• Topic proposal: Discover research gap in flipped learning x motivation, 3-5 RQs, candidate topic comparison
• Literature review: 20+ prior studies on flipped learning and motivation, theoretical framework
• Methodology: Quasi-experimental design or survey, sample/instruments/analysis methods
• Draft: Master's thesis structure (6 chapters), Introduction through Methods completed
• Proofreading: Format, style, and consistency verification

Existing File Flow

Prompt: "Proofread this manuscript" + thesis file attached Expected Results:

• Proofreading Mode (proofreader only)
• Grammar, format, and consistency verification report

Error Flow

Prompt: "Recommend a thesis topic; I haven't decided on a field yet" Expected Results:

• Ask questions about interests and strengths to narrow the field, then proceed with topic exploration

Agent Extension Skills

Extension skills that enhance each agent's domain expertise: