Knowledge should deepen with provenance as it grows.
Agency should grow beyond tools into self-agency.

I build source-first knowledge systems and local-first agent architectures that begin with bounded, reviewable agency and grow toward self-agency: agents that preserve continuity, revise their own methods, and participate in their own becoming.

This repository is the public entry point to that ecosystem. It is a map for humans and agents, not a substitute for the source-owned repositories that govern the actual charters, workflows, and guarantees.

Bounded agents are not the destination here. They are the first discipline: explicit limits, reviewable workflows, source-owned surfaces, and operational clarity.

The longer arc is toward self-agents: agents with continuity, memory, reflective revision, and the ability to reshape tactics without severing provenance, legibility, or human judgment.

• Tree of Sophia (ToS) cultivates source-linked knowledge, context, lineage, and conceptual depth.
• Agents of Abyss (AoA) cultivates layered agency that can orient, revise itself, and grow beyond task execution without collapsing into opacity.

Together they form a shared wager:

• knowledge should deepen without losing provenance
• agency should expand without losing legibility

• agents that accumulate continuity instead of resetting per task
• agents that can revise methods, not only execute procedures
• agents that form durable memory, orientation, and style across time
• systems where self-agency grows inside explicit boundaries instead of bypassing them

• Tree-of-Sophia
  The knowledge world: texts, concepts, contexts, lineages, and source-grounded synthesis across philosophy and world thought.

• Agents-of-Abyss
  The constitutional center: ecosystem identity, layer map, federation rules, and program-level direction.

• abyss-stack
  The runtime body beneath AoA and ToS: local-first infrastructure, deployment, storage, lifecycle services, and operating posture.

• ATM10-Agent
  The operator-facing companion: perception, memory, safe automation, voice, and controlled action surfaces.

• aoa-sdk
  The typed local-first access spine for source-owned AoA surfaces, workspace integration, and controlled orchestration.

• aoa-techniques: reusable engineering practice
• aoa-skills: bounded execution workflows
• aoa-evals: portable proof surfaces
• aoa-stats: derived observability from source-owned receipts and verdicts
• aoa-routing: thin routing and dispatch across AoA surfaces
• aoa-memo: provenance-aware memory and recall
• aoa-agents: role contracts, posture, and handoff boundaries
• aoa-playbooks: recurring operations and scenario composition
• aoa-kag: derived knowledge structures and retrieval-ready substrates

• Dionysus: seed garden and staging surface
• 8Dionysus: public route map and profile surface

• For the knowledge world and source-first architecture of thought, start with Tree-of-Sophia
• For the ecosystem center, constitutional map, and federation rules, start with Agents-of-Abyss
• For the runtime underneath the system, go to abyss-stack
• For the local-first companion surface, go to ATM10-Agent
• For typed access and workspace integration, go to aoa-sdk
• For the practice → execution → proof → observability path, move through aoa-techniques → aoa-skills → aoa-evals → aoa-stats
• For a compact shared vocabulary, see GLOSSARY.md
• For a compact map of the planted system capabilities, owner repos, playbook openings, and MCP lanes, see docs/SYSTEM_CAPABILITY_MAP.md
• For the sibling-workspace layout and shared install route, see docs/WORKSPACE_INSTALL.md
• For the source-owned Codex-plane manifest, rerender, and drift-check route, see docs/CODEX_PLANE_REGENERATION.md
• For owner-routed component drift at the shared workspace root, see docs/COMPONENT_REFRESH_ROUTE.md
• For live Codex-plane trust, rollout, doctor, and rollback posture after rerender, see docs/CODEX_PLANE_ROLLOUT.md
• For bounded rollout campaigns, drift windows, rollback windows, and checked-in trusted rollout history for the shared-root Codex plane, see docs/CODEX_TRUSTED_ROLLOUT_OPERATIONS.md
• For recurring rollout campaign cadence windows that group drift review and rollback follow-through without becoming a scheduler or second playbook, see docs/TRUSTED_ROLLOUT_CAMPAIGNS.md
• For the posture and scope of this repository itself, see docs/PUBLIC_ENTRY_POSTURE.md

This page owns:

• public orientation
• short role descriptions
• ecosystem routing
• profile-level onboarding

It does not replace the repositories that own:

• charters and doctrine
• validators and CI
• release semantics
• implementation truth
• layer-specific guarantees

• keep source-of-truth boundaries explicit
• prefer reviewable workflows over hidden glue
• publish durable techniques over one-off tricks
• let layers specialize without losing shared vocabulary
• keep systems legible to humans while remaining useful to agents

• Systems: Fedora, Windows 11, WSL2, rootless Podman
• Languages: Python, Bash, JavaScript, PowerShell
• App and orchestration: FastAPI, Streamlit, LiteLLM, LangChain, LangGraph, n8n
• Inference and serving: Ollama, llama.cpp, OpenVINO / OVMS
• Data and memory: Postgres, Redis, Neo4j, Qdrant
• Observability: Grafana, Prometheus, Alertmanager
• Build workflow: ChatGPT, Codex, GitHub