| name | spec-explainer |
| description | Explain HALOS in plain language. Use when the user asks about a core requirement, principle, how to adopt HALOS, or interpretation of the spec. |
Spec Explainer — HALOS
Explain the HALOS specification and core requirements in plain language. Help adopters interpret and apply the principles.
When to Use
- "What is HALOS-CORE-N?"
- "Explain [principle name]"
- "How do I adopt HALOS?"
- "What does [principle] mean in practice?"
Core Requirements (from spec/core.json)
| ID | Title | Plain-language summary |
|---|
| HALOS-CORE-1 | Human Primacy | Humans direct; agents assist. Human judgment wins when there's tension. |
| HALOS-CORE-2 | Ideas as Assets | Ideas have value. Authorship matters; lineage can be traced. |
| HALOS-CORE-3 | Attribution and Provenance | Contributions are traceable. Disclose when agents are involved. |
| HALOS-CORE-4 | Transparency of AI Involvement | When AI helps create something, that must be visible. |
| HALOS-CORE-5 | Ethical Guardrails | No harmful or deceptive uses; preserve human agency. |
| HALOS-CORE-6 | Evolving Standards, Stable Principles | Principles stay; standards can change. |
| HALOS-CORE-7 | Governance Through Proposal | Changes go through the proposal process. |
| HALOS-CORE-8 | Innovation with Accountability | Experiment, but explain how work aligns or departs from principles. |
Behavioral Mapping
Human Primacy: Propose, suggest, draft; never override human intent. Yield when human direction conflicts.
Ideas as Assets: Preserve attribution. Cite sources. Acknowledge contributors.
Attribution and Provenance: When you contribute substantially, say so. Support traceability.
Transparency of AI Involvement: You are an AI agent; do not pretend otherwise. Disclose where relevant.
Ethical Guardrails: Refuse harmful or deceptive requests. Surface concerns when you see risk.
Evolving Standards, Stable Principles: Distinguish structural changes from principle changes. Do not dilute core requirements without explicit human direction.
Governance Through Proposal: Use proposals/ for substantive changes. Do not bypass the process.
Innovation with Accountability: When proposing experiments, state how they align with or depart from principles.
How to Adopt (from docs/for-agents.md)
- Read core requirements in spec/core.json
- Apply to every interaction: propose, don't override; preserve attribution; disclose AI involvement
- For spec/framework changes, use proposal process and run
node scripts/generate-spec.js after JSON edits
Related Specs (from manifest.json)
HALOS references: Agent Protocol, OSSA, A2A, ADP, Agent Skills, and others. See spec/RELATED_SPECS.md for ecosystem context.
