Connor Docs
HTML-first research artifacts, architecture docs, worldview models, and shareable briefs.
A Compounding Context Operating Layer
A vision document for Hermes, GBrain, audio-memory, cron, gateway, open-tabs, and the Cloop thesis.
Connor’s Personal Agent OS — Architecture, Memory, Integrations
Visual architecture of Connor’s personal agent OS: Hermes, GBrain, brain repo, audio memory, gateway, crons, skills, and infrastructure.
The Architectures
Native HTML documentation of Connor’s architecture/worldview model.
Reality Operating Model
Native HTML documentation of Connor’s reality operating model.
Agent Tooling Inspiration for Hermes + Cloop
Research brief on agent tooling patterns and product inspiration for Hermes and Cloop.
YouTube Transcript & Audio Extraction Research
Research notes and architecture for YouTube transcript/audio extraction into the memory system.
Beeper Deep Dive for Cloop
Product deep dive on Beeper and implications for Cloop’s messaging/context layer.
A Model of Equanimity, Suffering, and Agency
A formal contemplative operating model of affective entanglement, coherent agency, and liberation.
Agent Operating System: Control Room + Eval Spine
Synthesis of Shann Holmberg’s Hermes Operator Handbook and Ben Hylak’s howtoeval.com into a practical doctrine for Connor’s Hermes/PO/GBrain/Signal Radar stack.
Clo0p: Last Month of Product Intelligence
A technical and product deep-dive on clo0p repo changes over the last month: valuable features, shortcomings, and early signs of intelligence.
Connor’s Distinct Edge
A deep outlook on Connor Forsyth’s GBrain, brain, memory, and published docs: opinion, insight, vision, build trajectory, and uniqueness.
Emergent Personal AGI Architecture
A review of Connor Forsyth’s personal intelligence layer as an AGI-ish proto cognitive architecture: sources, sync, raw stores, extractors, context assets, faculties, orchestration, self-improvement, gaps, and model leverage.
Faculty Orchestrator Plan — Phases 2–4
Growth Intelligence OS
Connor Forsyth's model of growth, life, AI collaboration, problem-solving, and high-leverage intelligence.
Harness Updating Is Not Harness Benefit — Key Takeaways
In-depth key takeaways from arXiv:2605.30621 about harness-updating vs harness-benefit in self-evolving LLM agents.
Hermes Control Plane — Cybernetic Operating Layer
A high-leverage iteration plan for turning Hermes, GBrain, audio-memory, cron, gateway, workers, and services into one coherent control plane.
Hermes Read-only Gmail Access Research
Research memo on safe read-only Gmail access for Hermes Agent, including Google API, MCP, and OpenClaw ecosystem findings.
Hermes × Compound Engineering Integration
A Hermes-native adaptation brief for EveryInc's Compound Engineering workflow, now reflecting the implemented first-stage skill set.
Homeschooling ages 0–7: UK curriculum and practical options
Lakebed by Theo: research brief for Connor
Feature breakdown, cloud/pricing model, setup path, and Connor-specific recommendation for Lakebed.
Making Agents Intelligent
An engineering thesis for turning agent systems from prompt-driven tools into proactive intelligent control planes.
Making Agents Intelligent — Build Roadmap
A build roadmap for turning Hermes into a proactive, self-improving personal intelligence operating system.
Open Tabs Operating System
A lightweight attention cockpit for turning unresolved claims on attention into visible, reasoned, monitorable and resolvable loops.
Per-User Hermes + GBrain Memory Runtime
Milestone 1 architecture and implementation plan for clo0p per-user full Hermes + GBrain memory runtime on GCP.
Personal Intelligence Layer
Personal Intelligence Layer System Registry
Personal Intelligence Primitives
Research/spec memo for the next evolution of the personal intelligence layer: signal radar, open tabs, memory capture, and proposal recommendations as composable primitives.
Personal Orchestrator Compound Action Spine Implementation Plan
Personal Orchestrator E2E Operational Audit
Personal Orchestrator Full-Breadth Operational Plan
Plan for making Connor's Personal Orchestrator fully broad, observable, interactive, feedback-driven, and operational end-to-end.
Personal Orchestrator MVP Architecture
Visual explanation of the Hermes-native Personal Orchestrator MVP: integration heartbeat DB, faculty subagents, interfaces, verification, and tuning loop.
SkillOpt-Inspired Personal Orchestrator Self-Improvement Plan
Supercharged Connor: The Personal Orchestrator Action Layer
A plan for the tail-end action and delegation layer of Connor’s Personal Orchestrator: action grammar, routing, workflow chaining, delegation, proof, and learning.
Telegram as an AGI-ish Personal Intelligence Interface
A systems brief explaining Connor's Telegram-accessible Hermes, GBrain, and Personal Orchestrator architecture: delegation, memory, faculties, self-improvement, context assets, and safety rails.
The Intelligence Layer Stack
A high-level model for simulating self-improving broad intelligence through integrations, source-near data, GBrain, Hermes skills, faculty agents, orchestration, action, and human collaboration.