{"id":48848905,"url":"https://github.com/abelgarza/nomos-relay","last_synced_at":"2026-04-15T07:01:58.465Z","repository":{"id":351245488,"uuid":"1210176863","full_name":"abelgarza/nomos-relay","owner":"abelgarza","description":"A cybernetic orchestration framework for local AI agents. Nomos Relay acts as a deterministic bridge for reasoning models (Gemma 4), enforcing linguistic cardinality and fail-closed   security profiles to transform stochastic thought into precise system actions. Built for local sovereignty, extreme token efficiency, and safe tool-calling.","archived":false,"fork":false,"pushed_at":"2026-04-14T08:25:26.000Z","size":29,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-04-14T08:33:15.313Z","etag":null,"topics":["ai-governance","automation","cybernetics","determinism","edge-computing","gemma4","ia-agen","language-model","local-ai","ollama","proxy","security"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/abelgarza.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-04-14T06:50:11.000Z","updated_at":"2026-04-14T08:25:30.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/abelgarza/nomos-relay","commit_stats":null,"previous_names":["abelgarza/nomos-relay"],"tags_count":6,"template":false,"template_full_name":null,"purl":"pkg:github/abelgarza/nomos-relay","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/abelgarza%2Fnomos-relay","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/abelgarza%2Fnomos-relay/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/abelgarza%2Fnomos-relay/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/abelgarza%2Fnomos-relay/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/abelgarza","download_url":"https://codeload.github.com/abelgarza/nomos-relay/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/abelgarza%2Fnomos-relay/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":31830157,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-14T18:05:02.291Z","status":"online","status_checked_at":"2026-04-15T02:00:06.175Z","response_time":63,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["ai-governance","automation","cybernetics","determinism","edge-computing","gemma4","ia-agen","language-model","local-ai","ollama","proxy","security"],"created_at":"2026-04-15T07:01:57.418Z","updated_at":"2026-04-15T07:01:58.458Z","avatar_url":"https://github.com/abelgarza.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Nomos Relay\n\n\u003cp align=\"center\"\u003e\n  \u003cimg src=\"https://img.shields.io/github/actions/workflow/status/abelgarza/nomos-relay/ci.yml?branch=main\u0026style=for-the-badge\u0026logo=github\" alt=\"CI Status\"\u003e\n  \u003cimg src=\"https://img.shields.io/github/license/abelgarza/nomos-relay?style=for-the-badge\" alt=\"License\"\u003e\n  \u003cimg src=\"https://img.shields.io/badge/Python-3.10+-3776AB?style=for-the-badge\u0026logo=python\u0026logoColor=white\" alt=\"Python Version\"\u003e\n  \u003cimg src=\"https://img.shields.io/badge/Ollama-Compatible-black?style=for-the-badge\u0026logo=ollama\" alt=\"Ollama\"\u003e\n\u003c/p\u003e\n\n---\n\n**Nomos Relay** is a cybernetic orchestration framework for local AI agents. It provides a deterministic bridge between stochastic reasoning models and secure system execution. \n\nInspired by the concept of *Nomos* (the ordering of space and law) and the *Caveman* philosophy of linguistic compression, this framework ensures that local agents (like Gemma 4) operate within strict safety boundaries while maintaining perfect structural integrity in their output.\n\n## Multi-Agent Architecture\n\nNomos Relay operates through a collaborative loop where each agent has a specific, \"constitutional\" role defined in its Modelfile:\n\n| Agent | Role | Responsibility |\n| :--- | :--- | :--- |\n| **Overlord** | Strategic | Decomposes high-level objectives into technical Kanban tasks. Maintains tech stack consistency. |\n| **Planner** | Tactical | Analyzes codebase context (RAG) and the current task to generate a compressed plan. |\n| **Relay** | Transducer | Converts plans into strict JSON/Bash commands. Enforces Fail-Closed safety. |\n| **Git** | Versioning | Specialized in analyzing diffs and generating Conventional Commits. |\n| **Cloud** | Observer | High-parameter model variant for remote/TPU-powered orchestration. |\n\n---\n\n## Installation\n\n### 1. Install Python Package\n```bash\n# Clone the repo and install in editable mode\npip install -e .\n```\n\n### 2. Build the Model Constitutions\nEnsure you have [Ollama](https://ollama.com/) installed and running.\n```bash\nnomos build\n```\n\n---\n\n## The Nomos CLI\n\nThe `nomos` command is the primary entry point for both interactive and autonomous development.\n\n### 1. Task Execution (One-Shot)\nRun a single task in a specific security profile:\n```bash\nnomos \"List all python files in src/\" --profile read-only\nnomos \"Create a basic API with FastAPI\" --profile developer --execute\n```\n\n### 2. Autonomous Architect Mode (`--auto`)\nThe most powerful mode. Nomos becomes an autonomous engineer that iterates until the goal is met.\n```bash\nnomos \"Build a full TUI dashboard based on README.md\" --auto --profile developer\n```\n**In this mode, Nomos will:**\n1.  **Recon:** Scan the environment to detect the language (Go, Python, etc).\n2.  **Plan:** Overlord creates a Kanban board in `.nomos/kanban.json`.\n3.  **Branch:** If a Git repo is detected, it moves to a safe `.nomos` branch.\n4.  **Execute:** Iterates through tasks, updating its memory (RAG) after each step.\n5.  **Commit:** Automatically commits successful tasks using the Git Agent.\n\n### 3. General Utility\n```bash\nnomos ask \"How do profiles work in this framework?\"  # Quick Q\u0026A with the Planner\nnomos list                                           # List available Nomos models\nnomos index --reset                                  # Wipe and rebuild RAG memory\n```\n\n---\n\n## Cognitive Memory (RAG)\n\nNomos uses **LanceDB** to maintain a deep semantic understanding of your codebase.\n\n- **Incremental Sync:** In `--auto` mode, Nomos automatically re-indexes changed files after each task, ensuring the next step is always context-aware.\n- **Manual Indexing:** Run `nomos index` to update the memory after manual changes.\n- **Hard Reset:** Use `nomos index --reset` to clear the vector store and start fresh.\n\nMemory state lives in the `.nomos/` directory of your workspace.\n\n---\n\n## Integration Examples (Extending the Framework)\n\nNomos Relay is designed to be a \"Cybernetic Core\" that can be used by other tools:\n\n### OpenAI-Compatible Proxy\nRunning `node proxy.js` starts a middleware that allows any OpenAI-compatible tool to use Nomos agents and RAG.\n\n### Implementation Case Studies:\n- **`oc-nomos.sh` (OpenCode):** Launches OpenCode using Nomos Relay as the backend, giving the editor full codebase awareness and execution safety.\n- **`pi-nomos.sh` (Pi):** Integrates the Pi agent with Nomos to provide a terminal-based cybernetic assistant.\n\n---\n\n## Security Profiles\n\nNomos enforces strict \"Fail-Closed\" governance through profiles:\n\n- **`read-only`**: (Default) Discovery only. No filesystem mutations allowed.\n- **`repo-safe`**: Allows `git`, `mkdir`, and `touch`. Safe for repo management.\n- **`developer`**: Full engineering access. Can use `pip`, `npm`, `python`, etc.\n\nEvery command is validated against a global `DENYLIST` to prevent system-level damage.\n\n---\n*Created by **Abel Garza Ramírez**. A Cybernetic approach to AI Agency.*\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fabelgarza%2Fnomos-relay","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fabelgarza%2Fnomos-relay","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fabelgarza%2Fnomos-relay/lists"}