For AI Agents

How HALOS works with AI agents — in this repo and in yours

HALOS defines principles for human-agent collaboration. This page explains how agents interact with HALOS in two contexts: working in this repository, and adopting HALOS elsewhere.

Working in This Repository

This repo is the HALOS community home — website, narrative, proposals, and governance. Agents working here must follow the HALOS contract.

Key files:

File	Purpose	Convention
AGENTS.md	Contract and agent definitions	Cursor, agent frameworks
.cursor/rules/	Cursor rules (halos-alignment, etc.)	Cursor
spec/manifest.json	Machine-readable spec discovery	OSSA, Agent Protocol, tooling

When working here:

1. Read the HALOS contract in AGENTS.md or .cursor/rules/halos-alignment.mdc
2. Apply the eight core principles to every interaction
3. For spec or framework changes, use the proposal process (proposals/)

Adopting HALOS in Your Repository

The canonical specifications and adoption toolkit live in halos-spec:

• Adoption guide — step-by-step for humans
• Agent prompt — agent-executable adoption process
• Templates — starter halos.yaml, adoption doc, conflict register
• Profile schema — validates your governance profile

Adoption has two phases:

1. Governance — create a halos.yaml profile, map existing artifacts, surface conflicts
2. Provenance — add .halos.json records, integrate with domain standards (SLSA, CycloneDX, etc.)

Links

• Principles
• Specification
• Agent Files — Rules, skills, and discovery files served on this site
• Governance