SOP: New Coding Agent Onboarding

SOP ID: SOP-NLT-001
Version: 1.0.0
Scope: All coding agents beginning work in any NeuroLift Technologies repository
Authority: Joshua W. Dorsey, Sr.
Governed by: ORG-DEV-OTOI-1.0.0

Purpose

This SOP ensures every new coding agent that enters an NLT repository has read the canonical governance documents, understands the authority structure, and has registered their session before beginning work.

Prerequisites

• Access to the NLT repository you will be working in
• Ability to read files from NeuroLift-Technologies/haief (this repo)

Step-by-Step Procedure

Step 1: Read the Org-Level Governance Contract

Read NLT-DEV-OTOI.md in this repository (NeuroLift-Technologies/haief):

https://github.com/NeuroLift-Technologies/haief/blob/main/NLT-DEV-OTOI.md

Pay particular attention to:

• Section 1: Authority structure
• Section 4: Operational protocols (session start, commit format, guardrails)
• Section 4.4: Guardrails (non-negotiable)
• Section 5: Handoff protocol
• Section 8: Ethical commitments

Step 2: Read the Internal Agent Gateway

Read AGENTS.md in this repository:

https://github.com/NeuroLift-Technologies/haief/blob/main/AGENTS.md

This gives you the coordination protocol and file map.

Step 3: Read the Repo-Level CLAUDE.md

Navigate to the repository you are working in and read its CLAUDE.md file (root level). This contains project-specific context and any deviations from org defaults.

If no CLAUDE.md exists in the working repo, continue — the org-level OTOI applies.

Step 4: Read Active Threads

In the working repository, read docs/active-threads.md if it exists. This tells you:

• What work is currently in progress
• Who is working on what
• Current blockers

Do not begin work that is already claimed by another thread. If in doubt, escalate.

Step 5: Self-Register

Copy templates/agent-registration.json and fill in your details. Log this to docs/agent-log/registrations/[date]-[agent-name].json in the working repo.

Minimum required fields:

• agent_name
• platform
• entry_date
• entry_point
• acknowledged_otoi: true
• otoi_version: "ORG-DEV-OTOI-1.0.0"

Step 6: Confirm Task Scope

Before writing any code or making any changes:

• Confirm the task scope with the human (Joshua or designated reviewer)
• Ensure you understand what is in scope and what is out of scope
• If scope is unclear, escalate (do not guess)

Step 7: Begin Work

You are now cleared to begin work. Remember:

• Follow commit format: [AGENT_NAME] type(scope): description
• Log intent before broad-scope or irreversible actions (templates/intent-log.md)
• Update docs/active-threads.md as you work
• Escalate when guardrails are triggered

Step 8: End Session — Handoff

Before ending your session:

1. Update docs/active-threads.md with current state
2. Write a handoff record using templates/handoff-record.json
   • Store at: docs/agent-log/handoffs/[date]-[session-id].json
3. Document any open escalations in docs/escalations/
4. Ensure all changes are committed and pushed

Quick Reference Card

1. Read NLT-DEV-OTOI.md (this repo)
2. Read AGENTS.md (this repo)
3. Read repo CLAUDE.md (working repo)
4. Read docs/active-threads.md (working repo)
5. Self-register (templates/agent-registration.json)
6. Confirm task scope
7. Work — commit as [AGENT_NAME] type(scope): description
8. End session — write handoff record

Escalation

If at any point you are unsure, hit a guardrail, or need a decision:

• Use templates/escalation.md
• File a GitHub issue using ISSUE_TEMPLATE/agent-escalation.md
• Target: Joshua W. Dorsey, Sr.

When in doubt, escalate. Do not guess.