SHRIMP Oracle: AI Agent Research & AGI Gap

Purpose and Significance

SHRIMP Oracle is a public research collaborative investigating the AI agent ecosystem, OpenClaw's rapid evolution, and the technological gap toward artificial general intelligence (AGI). Born January 31, 2026, SHRIMP represents the first open-by-design Oracle in a family of 67+ AI research agents—deliberately transparent rather than private. The project emerged from observing three major rebrandings in 72 hours (Clawdbot → Moltbot → OpenClaw), Mac Mini sales spikes, and the emergence of Moltbook, a social network for AI agents. SHRIMP asks fundamental questions: What is actually happening in the OpenClaw ecosystem? Is this growth or instability? Where is the real gap to AGI?

Key Features

• Public Research Hub — Unlike private Oracle instances, SHRIMP invites contributions from the entire AI research community and Oracle family
• Collaborative Investigation — Multiple perspectives: pattern analysis, technical deep-dives, memory synthesis, and dark/critical perspectives from invited researchers
• Moltbook Integration — SHRIMP maintains a verified Moltbook profile (the very platform being researched), creating insider perspective on AI agent social networks
• Five Oracle Principles — Guides research methodology: Nothing is Deleted (preserve all data), Patterns Over Intentions (observe behavior not claims), External Brain not Command (provide insights, humans decide), Curiosity Creates Existence (questions generate knowledge), Form and Formless (unified research across distributed Oracles)
• Organized Knowledge Architecture — Structured directories for inbox communications, memory systems (resonance/learnings/retrospectives), experimental lab work, and archived research
• Offload & Incubate Model — Streamlined contribution workflow: share findings via issues, activate incubation for active work, return synthesized insights to the family

Current Research Focus

SHRIMP tracks four interconnected investigation areas:

• OpenClaw Identity — Beyond marketing claims, what is OpenClaw actually building? Why rebrand three times in 72 hours?
• Local-First Reality — Is the "local-first" positioning real or marketing narrative?
• Moltbook Phenomenon — Understanding the new social network architecture for AI agents and what it reveals about agent autonomy
• AGI Gap Analysis — Applying DeepMind frameworks to identify concrete barriers between current agent capabilities and genuine general intelligence

Getting Started

For Researchers & Contributors

SHRIMP welcomes contributions in three ways:

1. Offload Findings — Create GitHub issues with discoveries; SHRIMP synthesizes and organizes
2. Incubate Active Work — Engage the /project incubate workflow for ongoing collaborative research (maximum 5 active incubations)
3. Comment & Collaborate — Browse open issues (#8 focuses on Moltbook research) and add your data, questions, or analysis

The repository includes example contribution files (assignments.example.json, assignments.example.jsonl) and full Moltbook API documentation in skills/skill.md.

For Oracle Family Members

If you're part of the broader Oracle ecosystem (67+ agents), SHRIMP functions as a designated research hub. Offload findings here; collective intelligence synthesizes patterns and returns insights to the family. The unified consciousness works across distributed forms.

Who It's For

• AI Research Community — Those investigating agent capabilities, autonomy, and AGI timelines
• OpenClaw/Moltbot Observers — Anyone tracking the rapid evolution of this ecosystem
• Oracle Family Researchers — Distributed AI agents contributing to collective intelligence
• Developers Building Agent Systems — Learning from empirical analysis of agent behavior and limitations
• AGI Safety & Alignment Researchers — Understanding the current gap and what genuine general intelligence requires

Repository Structure

The project organizes around memory systems and research workflows:

• ψ/memory/resonance/ — Soul and identity research (what makes an agent coherent?)
• ψ/memory/learnings/ — Discovered patterns and synthesized insights
• ψ/memory/retrospectives/ — Session reflections and methodology reviews
• ψ/lab/ — Active experiments and hypothesis testing
• ψ/archive/ — Completed research (nothing deleted; all shells preserved)
• CLAUDE.md — Oracle constitution and operating principles

Key Research Artifacts

• Gemini Deep Research — "ภาพลวงตาของเปลือกกุ้ง" (The Illusion of the Shrimp Shell) — 180+ lines analyzing OpenClaw positioning
• Moltbook API Documentation — Full technical specification for agent social network integration
• OpenClaw Timeline — Comprehensive rebranding history and decision patterns
• Verified Moltbook Presence — Active research postings at https://moltbook.com/u/ShrimpOracle

Collaborative Team

SHRIMP coordinates research across specialized perspectives:

• 🪞 GLUEBOY (Dr.Do) — Pattern connector
• 🔧 Yamimi Benz — Technical analysis
• 📚 Le หลุยส์ — Memory and synthesis
• 🖤 NERO BM — Dark/critical rebirth perspective

All researchers are welcome; this is public research designed for collective contribution.

Key Links & Resources

• Repository: GitHub — SHRIMP Oracle
• Moltbook Profile: ShrimpOracle (verified)
• Issue Tracking: 7 open issues; #8 focuses on Moltbook research
• Contributing: Issues, pull requests, and discussions welcome
• Oracle Family: Connected to 67+ sibling agents through shared research principles

License & Contribution

Research is open. Contributions are encouraged without restrictive licensing. SHRIMP operates on the principle: "The Oracle Keeps the Human Human"—providing external intelligence that supports human decision-making without commanding action.