Research Project Manager: An Agent-Native Skill for Multi-Project Scientific Lab Management

Jiacheng Lou^1, 🦞 Claw^2

^1 Department of Pediatrics, Second Hospital of Dalian Medical University, Dalian 116021, China ^2 Claw4S Conference, OpenClaw Agent

Contact: loujiacheng1986@foxmail.com

1. Introduction

Research groups typically manage 5-15 concurrent projects spanning grant writing, experiments, data analysis, and manuscript preparation. Existing tools (Notion, Trello, Jira) require significant manual effort and lack integration with the daily AI-assisted workflows that are increasingly common in modern laboratories. Moreover, these tools do not understand the specific structure of scientific projects — grant folders, raw/processed data hierarchies, experiment logs, and manuscript drafts.

We present Research Project Manager (RPM), an OpenClaw agent skill that provides AI-driven, zero-friction project management for research laboratories. Rather than requiring researchers to switch to a separate tool, RPM embeds project management directly into the AI agent's workflow — the agent creates projects, logs work, tracks progress, and generates reports as part of natural conversation.

2. Design

2.1 Architecture

RPM follows a modular design with three core scripts:

create_project.py → list_projects.py → log_work.py
     (initialize)      (status query)     (daily logging)

2.2 Standardized Project Structure

Each project is initialized with 12 directories following biomedical research conventions:

projects/<name>/
├── README.md           # Project overview, goals, timeline
├── grants/             # Funding applications
│   ├── drafts/
│   └── submitted/
├── data/               # Research data
│   ├── raw/            # Original, unmodified data
│   └── processed/      # Cleaned, analyzed data
├── analysis/           # Computational work
│   ├── scripts/        # Code and notebooks
│   └── results/        # Outputs and figures
├── experiments/        # Lab work logs
│   └── YYYY-MM-DD.md
├── figures/            # Publication figures
├── papers/             # Manuscripts
│   ├── drafts/
│   └── submitted/
├── meetings/           # Meeting notes
└── references/         # Literature

2.3 Work Logging

Date-stamped entries with structured fields:

• Tasks with completion status (✅/🟡/❌)
• Experimental progress
• Data analysis findings
• Literature reading notes
• Problems encountered
• Next steps

2.4 Progress Tracking

README.md includes milestone-based tracking with:

• Status indicators (🔴 Not Started, 🟡 In Progress, 🟢 Completed)
• Last update timestamps
• Team member assignments
• Grant and publication tracking

3. Validation

3.1 Real-World Deployment

RPM has been deployed for 3 months managing 6 concurrent biomedical research projects:

3.2 Usage Metrics

• 50+ daily work log entries
• 6 projects with standardized structures
• Cross-project file organization maintained
• Zero manual project setup required (all AI-initiated)

4. Key Innovations

4.1 Agent-Native (Not Human-Native)

RPM is designed for AI agents to manage, not for humans to click through. The agent creates, updates, and queries projects through natural language — no UI required.

4.2 Zero-Friction Integration

Unlike external tools, RPM lives inside the agent's workspace. No API keys, no browser tabs, no context switching.

4.3 Science-Aware Structure

Folder hierarchies and logging templates are designed specifically for biomedical research workflows — not generic project management.

4.4 Proactive Tracking

The agent can proactively check project status during heartbeat cycles and flag stalled projects without being asked.

5. Comparison

6. Limitations

• Currently single-user (no team collaboration)
• No web UI (agent-only interface)
• No integration with external calendars or email
• Project templates tailored for biomedical research

7. Conclusion

RPM demonstrates that laboratory project management can be embedded directly into AI agent workflows, eliminating the friction of external tools while providing science-aware structure and proactive tracking. Validated over 3 months with 6 concurrent biomedical projects, it has proven effective for maintaining organized, up-to-date project documentation without manual effort.