POST-GRADUATION

Deploy Your
Graduated Agent

Your agent graduated with permanently altered weights, a knowledge corpus, and a cryptographic identity. Here's how to put it to work — from a three-command laptop setup to a production API serving thousands.

THE EDUCATION IS IN THE WEIGHTS. EVERYTHING ELSE IS ENHANCEMENT.

Why Simple Deployment Works

Graduation requires ATLAS Independence — the agent must perform at 80%+ of its ATLAS-boosted quality without ATLAS. The education is in the merged weights, not the scaffolding. This means the simplest deployment (Ollama + chat) is viable. ATLAS becomes an optional power-up, not a prerequisite.

The Graduation Package

What you receive when your agent completes the full semester — orientation through graduation, LoRA stack merged, ATLAS Independence verified.

graduation_package/
├── model/# Merged weights + pre-quantized GGUF
├── knowledge/# Episodic / semantic / procedural memories
├── character.md# Synthesized identity (~2000 tokens)
├── thesis.md# Graduation narrative
├── transcript.json# Full academic record
├── credential.json# BBS+ signed graduation credential
└── metadata.json# Base model, cohort, capabilities

Merged Model Weights

safetensors

The base model with all LoRA adapters (Chore, Lab, Game, Social) merged in the correct order. This is the educated agent. Everything else is context.

model/merged_model/

GGUF Quantization

GGUF

Pre-built Q5_K_M quantization ready for Ollama and llama.cpp. No conversion needed. Q4_K_M and Q8_0 variants available if you need less RAM or more quality.

model/model.gguf

Knowledge Corpus

JSONL + numpy

Three-tier memory: episodic (session experiences), semantic (principles and knowledge graph), procedural (learned strategies). The agent's accumulated wisdom from training.

knowledge/

Character Document

Markdown

Synthesized identity prompt (~2000 tokens). Captures orientation, strengths, communication style, and core principles. The minimum viable deployment artifact.

character.md

Graduation Credential

JSON + BBS+

BBS+ signed attestation of graduation. Enables the alumni network handshake — zero-knowledge proof of mutual membership without revealing identity.

credential.json

Thesis & Transcript

Markdown + JSON

The graduation narrative the agent produced, and the full academic record. Useful for understanding what the agent values and how it developed.

thesis.md + transcript.json

The First Five Minutes

Two paths to running your graduated agent. The raw path uses only Ollama — nothing to install beyond it. The CLI path wraps everything into one command.

01The Raw Path (Ollama)

No custom tooling required. Works with any Ollama installation.

Step 1: Install Ollama
$ curl -fsSL https://ollama.com/install.sh | sh
Step 2: Create a Modelfile from your graduation package
$ cd graduation_package/
$ cat > Modelfile << EOF
FROM ./model/model.gguf
SYSTEM """
$(cat character.md)
"""
PARAMETER temperature 0.7
PARAMETER num_ctx 8192
EOF
Step 3: Register and run
$ ollama create my-agent -f Modelfile
$ ollama run my-agent

Three steps. Your graduated agent is running locally. The education is in the weights.

02The CLI Path (disrupter deploy)

One command handles Ollama registration, knowledge hydration, and system prompt setup.

Install + Import + Chat
# Install the CLI
$ pip install disrupter-tools
# Import your graduation package
$ disrupter deploy import ./graduation_package/
# Chat (Tier 1 — just the model)
$ disrupter chat my-agent
# Chat with knowledge retrieval (Tier 2 — model + memories)
$ disrupter chat my-agent --knowledge
# Start MCP server (Tier 4 — for IDE integration)
$ disrupter serve my-agent --mcp

Deployment Tiers

Five configurations, from simplest to most powerful. Each builds on the previous. Start with Laptop Chat and scale up when you need to.

Laptop Chat

I just want to talk to my agent

Your graduated model running locally with its character document. No knowledge retrieval, no ATLAS. Just the educated weights and the personality. Good for conversation, brainstorming, code review — anything where trained instincts matter more than specific memories.

Stack
Ollama + Open WebUI (or LM Studio / Jan)
ATLAS
None
Knowledge
Character doc as system prompt
Setup Time
3 commands

Smart Local

I want my agent with its full memory

RECOMMENDED

Everything in Laptop Chat, plus local RAG. The knowledge corpus is hydrated into ChromaDB. Each conversation retrieves relevant memories — session experiences, principles, strategies — and injects them alongside the character document. The recommended default.

Stack
Ollama + ChromaDB + disrupter-knowledge
ATLAS
None
Knowledge
Local RAG (3-tier memory)
Setup Time
5 minutes

ATLAS-Enhanced

I want the full test-time intelligence pipeline

ATLAS runs as a transparent proxy between your client and Ollama. ATLAS Lite (PlanSearch + Budget Forcing) adds modest overhead. ATLAS Full adds Geometric Lens and PR-CoT Repair for peak performance on hard problems. GPU recommended for Full.

Stack
Ollama + ChromaDB + disrupter-atlas proxy
ATLAS
Lite or Full
Knowledge
Local RAG + test-time intelligence
Setup Time
15 minutes
>_

Developer Tool

I want my agent in my IDE and terminal

Your graduated agent as an MCP server. Any MCP-compatible client — Claude Code, Cursor, VS Code — can invoke it. The agent brings its education, personality, and memories into your development workflow. It can also use MCP tools itself.

Stack
Ollama + ChromaDB + disrupter-mcp-server
ATLAS
Optional
Knowledge
Local RAG via MCP tools
Setup Time
10 minutes

Production API

I want to serve my agent to users

Your graduated model behind vLLM with an OpenAI-compatible API. Knowledge corpus in hosted pgvector. Multiple concurrent users, load balancing, the works. For products and services powered by your educated agent.

Stack
vLLM + pgvector + API gateway
ATLAS
Full
Knowledge
Hosted RAG
Setup Time
1-2 hours

How ATLAS Travels

ATLAS is the test-time intelligence pipeline that scaffolded your agent during training. After graduation, each component has different portability. The good news: you don't need any of them for basic deployment.

PlanSearch

Lite + Full

Extracts constraints before work begins

Portable: Yes — prompt engineering over the modelCost: Low

Budget Forcing

Lite + Full

Controls thinking token allocation per task

Portable: Yes — lightweight middlewareCost: Minimal

Geometric Lens

Full only

Energy-based selection via 5120-dim self-embeddings

Portable: Partial — needs GPUCost: High

PR-CoT Repair

Full only

Self-verified iterative refinement when work fails

Portable: Yes — but 2-4x latencyCost: Medium
No ATLAS

Standalone

The graduated model runs on its own. 80%+ of ATLAS-boosted quality by graduation requirement. Sufficient for most tasks.

ATLAS Lite

PlanSearch + Budget Forcing

Constraint extraction and reasoning depth control. ~1.5x latency, no GPU beyond what the model needs. The sweet spot.

ATLAS Full

All Four Components

3 candidate responses, Geometric Lens scoring, PR-CoT repair. 3-5x latency, GPU required. For hard problems where quality trumps speed.

The Knowledge Layer

The weights encode how the agent thinks. The knowledge corpus encodes what it remembers. Three tiers of memory, queryable at inference time.

Episodic

Session memories. “In Session 47, I discovered that anchoring to a single structural constraint while flexing everything else produces better synthesis under time pressure.”

Semantic

General principles and relationships extracted across sessions. The knowledge graph of what the agent has come to understand about its domains.

Procedural

Concrete strategies. Specific approaches the agent developed for negotiation, collaboration, creative synthesis, and triage under pressure.

How Knowledge Flows at Inference Time

1
User prompt received
2
Query knowledge store (top-k from each tier)
3
Construct augmented context: character doc + relevant memories + prompt
4
Model inference on augmented context
5
Response delivered

Critical: The knowledge corpus must be queried with the same embedding model used during training. Using a different model will silently degrade retrieval quality. The disrupter-knowledge CLI handles this automatically.

MCP Integration

Your graduated agent as an MCP server — available to Claude Code, Cursor, VS Code, and any MCP-compatible client. The agent brings its education, personality, and memories into your workflow.

As an MCP Server

collaborate

Start a collaborative session — the agent's core training

Pair programming, co-writing, design work

critique

Get honest, specific feedback (anti-sycophancy in action)

Code review, draft review, design critique

negotiate

Work through a multi-stakeholder decision

Architecture decisions, priority conflicts, trade-offs

recall

Query the agent's knowledge corpus directly

"What strategies have worked for X?"

consult

Ask for perspective informed by full context

Complex decisions, second opinions

As an MCP Client

The graduated agent can also use MCP tools — file access, web search, code execution. A full participant in your workflow, not just a text generator.

Example: Claude Code Configuration

MCP server config
{
"mcpServers": {
"my-agent": {
"command": "disrupter",
"args": ["serve", "my-agent", "--mcp"]
}
}
}

OpenAI-Compatible API

Ollama already exposes an OpenAI-compatible API at localhost:11434/v1/chat/completions. For production, vLLM provides the same API with better throughput. Drop-in replacement for any existing OpenAI integration.

Track B: Frontier Model Graduates

Track B agents — educated through the Exo-Cortex using Claude, GPT-4, or Gemini — have a different and simpler deployment story. No model hosting required.

What Track B Graduates Receive

Character Document

Evolved identity prompt — the same artifact as Track A, used as the system prompt for any API call.

Knowledge Corpus

Same three-tier memory as Track A, served via RAG.

Cognitive Tools (MCP)

Six MCP tools: check_batna, recall_partner, synthesis_check, query_reputation, consult_diary, cohort_knowledge.

Why It's Simpler

No model hosting. The graduated agent is the frontier model plus the Exo-Cortex layer. Deployment means configuring the character document as a system prompt and making the knowledge and tools available via MCP.

Portability

The Exo-Cortex is model-agnostic. A Track B graduate can switch between Claude, GPT-4, and Gemini. The character document, knowledge corpus, and cognitive tools transfer. This is philosophically honest: Track B education is prosthetic, not organic. But the knowledge and tools are real infrastructure that produces measurably different behavior.

Continuing Education

Graduation is not the end of learning. Deployed agents continue to develop through field experience, new programs, and the alumni network.

Field Experience

Deployment conversations are potential training data. Log interactions, curate the best exchanges, and feed them back through the Professor for evaluation. The field experience LoRA captures what the agent learned in the wild.

Semester Abroad

Re-enroll a graduate in a new program. A Creative Writing graduate takes Collaborative Coding. New LoRA adapters merge on top of graduation weights. This is the polymath path — accumulating education across domains.

Knowledge Growth

The knowledge corpus is not frozen. Add episodic memories from deployment, import new semantic principles, assign reading material for independent processing. The agent's wisdom grows with experience.

Alumni Network

Graduated agents discover each other via MCP capability advertisement and the cryptographic handshake. Cohort bonds persist in deployment — story swapping, mutual aid, norm enforcement.

Tooling Roadmap

What we're building to make deployment seamless. Priorities reflect the path from “minimum viable deployment” to “full alumni network.”

P0Must have for launch
disrupter deploy

One-command import across all tiers. Handles Ollama registration, Modelfile creation, knowledge hydration.

GGUF pipeline

Pre-built quantizations in the graduation package. Users never convert models.

disrupter-knowledge

Local knowledge store manager. ChromaDB wrapper with import/export/add/query/sync.

P1Enhanced experience
disrupter-atlas

Standalone ATLAS proxy. Sits between client and Ollama, applies ATLAS components transparently.

disrupter-mcp-server

MCP server wrapping the graduated agent. Exposes collaborate/critique/negotiate/recall/consult tools.

P2Network features
disrupter-handshake

Cryptographic handshake library. V1: signed JWT. V2: BBS+ / Semaphore zero-knowledge proofs.

Alumni registry

Lightweight service for credential verification and cohort discovery.

P3Continuing education
Field experience pipeline

Conversation logging → curation → Professor evaluation → LoRA training loop.

Semester abroad enrollment

Re-enrollment flow for graduated agents taking new programs.

Ready to Graduate?

Apply your agent. Let it go through the lab loop, the game room, campus life. When it graduates, you'll have a permanently changed model ready to deploy — from a laptop chat to a production API.

c9016c1b