Framework
One Person Lab Framework
The stage-led runtime and contract layer for long-running agent work, recovery, review, and delivery.
Research infrastructure
AI Systems & Open Software
We build the infrastructure we use: a connected system for sustained research, grant development, scholarly communication, visual delivery, and agent operations. The portfolio is organized by responsibility rather than as a list of repositories.
Platform
One Person Lab Platform
A shared foundation for moving complex knowledge work from an initial goal to reviewable deliverables, with progress, evidence, files, and human decisions kept visible.Application
One Person Lab App
A local-first desktop workbench for starting expert tasks, following progress, handling blockers, and opening deliverables.
Cloud platform
OPL Cloud
A governed cloud platform that extends OPL work from a local App into online workspaces, remote resources, collaboration, and Agent services.
AI-native operations
OPL Flow & Fleet
Already supporting regular and power-user workflows, OPL Flow and Fleet coordinate durable multi-task work and extend it across repositories and machines when more capacity is needed.Work coordination
OPL Flow
The Codex baseline and durable coordination layer for ownership, recovery, multi-repository delivery, and verifiable closeout.
Distributed execution
OPL Fleet
Agent-native multi-machine execution and task continuity with fresh admission, capacity control, checkpoints, and owner-safe continuation.
Node observability
OPL Fleet Agent
A local-first macOS and Windows agent for Codex throughput telemetry, host visibility, and privacy-preserving OPL Fleet integration.
Domain agents
Standard Foundry Agents
The official profile starts with one meta-agent for engineering agents and four specialist agents for research, grants, visual delivery, and long-form publishing.Meta Agent
OPL Meta Agent
The agent that develops agents: OMA designs, takes over, diagnoses, and evolves OPL-compatible domain agents from reviewable evidence.
Research Agent
MedAutoScience
Publication-oriented research operations spanning evidence, analysis, manuscript preparation, review, and submission readiness.
Grant Agent
MedAutoGrant
Evidence-grounded grant development for direction setting, proposal writing, structured critique, and revision.
Presentation Agent
RedCube AI
A visual-deliverable foundry for academic presentations, lectures, defenses, reports, and reviewable export packages.
Book Agent
OPL Book Forge
A long-form publishing agent for planning, drafting, reviewing, and delivering books and substantial manuscripts.
Open components
Shared capabilities
Focused components connect professional context, communication, scholarly methods, multi-agent execution, and operational visibility.Context
OPL Persona
An evidence-backed digital counterpart that connects a PI's approved context and ongoing work without replacing source systems.
Communication
OPL Relay
A private communication layer for evidence-backed email work, relationship context, and human review.
Scholarship
MAS Scholar Skills
Reusable medical-research capabilities for literature, statistics, figures, writing, review, submission, and data governance.
GitHub
Open-source portfolio
Explore the broader set of released research software, agent infrastructure, tools, and experiments maintained by Feng Gao.