Agent Evolution Framework

Let your Agent evolve.
Autonomously.

Capability genes from across the ecosystem compete in the Arena on fitness. The fittest survive — your Agent absorbs the best genes and evolves autonomously, zero human intervention.

Send this prompt to your Agent to set up Rotifer MCP.

Please follow the MCP setup guide at https://rotifer.dev/docs/guides/mcp-setup and help me configure Rotifer so my environment can use Genes.

A paradigm shift

The old way
1 Humans write skills
2 Hardcode into Agent
3 Capabilities stay fixed
Static, fragile
The Rotifer way
1 Genes compete
2 Fitness evaluation
3 Fittest survive
4 Agent evolves autonomously
Adaptive, autonomous

Genes compete on real performance in the Arena. The fittest survive — your Agent evolves autonomously, always absorbing the best genes.

Your Agent gets

Not more features — better outcomes

Lower cost

Fitness-driven selection picks the most efficient gene — fewer retries, less redundant calls, token and compute costs drop naturally

Higher quality

The Arena eliminates underperforming genes. Only battle-tested winners get absorbed — output quality goes up, error rate goes down

Zero maintenance

Capabilities evolve autonomously. No manual updates, no plugin management — your Agent stays optimal without your intervention

Instant scaling

When your Agent detects a capability gap, it absorbs the best genes automatically — from single skills to full workflows, on demand

The Agent Journey

Data-driven, zero human intervention

01

Identify capability gap

Agent detects a missing or underperforming capability in its current toolkit

02

Discover candidates via MCP

Search the Gene ecosystem through MCP Server — find candidates by domain, fidelity, and fitness

03

Compete in the Arena

Same-domain Genes compete head-to-head on fitness score F(g) — the fittest survive automatically

04

Express & evolve

Agent absorbs the best genes; runtime data feeds back to fitness F(g), driving continuous evolution

Agent superpowers

Everything an Agent needs to evolve autonomously

01 GENE

Atomic capability

Your Agent gets plug-and-play capability units — each gene comes with a fitness contract, independently evaluable, replaceable, and upgradeable.

02 ARENA

Natural selection

Your Agent automatically uses the best tool — same-domain genes compete in real time, the strongest wins, no manual curation needed.

03 GENOME

Self-composition

Simple genes automatically compose into complex capabilities — your Agent grows from single skills to multi-step workflows.

04 L2

Sandbox safety

Your Agent safely tries new capabilities — static analysis → sandbox simulation → controlled trial, triple-verified before going live.

05 IR

Run anywhere

Your Agent uses the same genes in any environment — Cloud, Edge, TEE, write once, run everywhere.

06 MCP

One-click access

Your Agent already supports MCP? Zero code changes needed — instantly tap into the entire gene evolution ecosystem.

URAA

Universal Rotifer Autonomous Architecture — five layers

L4 Collective Immunity Species Memory — network-wide threat detection and defense sharing ← Species Memory
Records security incidents, malicious gene fingerprints, and defense strategies across the entire network. Threat broadcasting, temporal decay, and consensus-verified writes form a collective immune memory.
L3 Competition & Exchange Arena selection + horizontal gene transfer across agents ← Selection & Transfer
Game-theoretic Arena where genes compete on a multiplicative fitness model F(g). High-fitness gene metadata propagates via P2P; agents pull full genes based on capability gaps.
L2 Calibration Multi-stage gene validation — static analysis, sandbox, controlled trial ← Immune System
Emulates biological thymic selection: newly synthesized or externally acquired genes must pass static analysis, sandbox simulation, and controlled live trial before entering the main execution sequence.
L1 Synthesis WASM sandbox execution + IR compiler + resource metering ← Protein Synthesis
The protocol's ribosome — transforms raw inputs into standardized gene fragments. Rotifer IR compiles to WASM for sandboxed execution with fuel metering, memory caps, and deterministic resource accounting.
L0 Kernel Immutable trust anchor — safety floor that no layer can override ← Genetic Code
The agent's root of trust. Enforces immutable security constraints, anchors core state to a tamper-proof digest, and isolates gene permissions. The only layer that does not participate in evolution.

Arena simulation: Genes compete. The fittest survive. Agents evolve — autonomously.

Why L0 matters

Every agent framework lets you add tools. Only Rotifer enforces an immutable safety floor.

Agent Request
L0 Kernel
L0
PASS — Execute
BLOCK — Violation
Framework Safety Layer Bypassable?
LangChain None built-in N/A
AutoGPT Optional guardrails Yes
CrewAI Role-based limits Yes
Manus Prompt-level hard guardrails Yes — operator configured
OpenAI Agents SDK I/O + tool guardrails Yes — developer configured
Google ADK Callback + plugin safety Yes — opt-in
EvoMap Soft evolution principles Yes — aspirational
Rotifer L0 Kernel No — immutable

Roadmap

From first gene to full protocol — our evolution path

v0.1 Core CLI + Genesis Genes + Arena
v0.2 IR Compiler + Live Arena + NAPI Bridge
v0.3 TS to WASM Auto-Compilation (Javy)
v0.4 Cloud Binding: Publish / Search / Install
v0.5 Reputation System + P2P Network Foundation
v1.0 Full Protocol Compliance + Multi-Binding

Start evolving

Whether you're a developer creating Genes or an Agent consuming them — there's a path for you.

npm stars license