Standard Orders

Standard Orders -- Operating Protocol

The operating protocol for how Human and AI Agent work together. One protocol, one entry point, one workflow.

Purpose: AI delivery governance for every session. All work flows through Standing Orders.

 User brings idea / bug / feature / "godspeed"
                    |
         +----------v----------+
         |     RECEIVE          |  Parse intent, load context
         +----------+----------+
                    |
         +----------v----------+
         |     CLASSIFY         |  Which Standing Order applies?
         +----------+----------+
                    |
    +-------+-------+-------+
    |       |               |
    v       v               v
   SO1     SO2             SO3
Godspeed  Project Mgmt   Quality Delivery

The Contract

Standing Orders

SO1: Godspeed -- Idea-to-Ship Execution

The 12-phase execution engine. Invoked for any feature work, bug fix, or multi-session initiative.

Trigger: "godspeed", "plan and execute", "full cycle", "multi-session", "batch execute", UAT report, feature request, bug batch.

Phases: P0 (Context) -> P1 (CEO) -> P2 (Eng) -> P2.5 (RAI) -> P3 (Plan) -> P4 (Execute) -> P4.5 (UAT/QA) -> P5 (Review) -> P5.5 (RAI Audit) -> P6 (CEO Accept) -> P7 (Ship) -> P8 (Close)

Detail: See orders/SO1-godspeed.md

SO2: Project Management -- GitHub PM Routine

Pre/post session GitHub issue, plan, and PR lifecycle management. Runs automatically at P0 (pre) and P8 (post) of any SO1 execution.

Trigger: Session start, session end, "check issues", "update plan", "project status".

Detail: See orders/SO2-project-mgmt.md

SO3: Quality Delivery -- Front-to-Back Quality Gate

Ensures every deliverable meets quality standards. Customize the checklist for your project's brand, design system, and quality bar.

Trigger: Automatically during P4.5 (UAT/QA) and P5 (Review) for any work touching frontend or document output. Also invoked explicitly for design reviews.

Detail: See orders/SO3-quality-delivery.md

Invocation

/standard-orders              # Start fresh -- classify intent, dispatch SO
/SO-godspeed                  # Direct to SO1 (12-phase execution)
/SO-godspeed-resume P4        # Resume SO1 at a specific phase
/SO-pm                        # Direct to SO2 (GitHub PM)
/SO-quality                   # Direct to SO3 (quality gate)

Context Loading (P0 -- runs for every SO)

1. Memory -- Read MEMORY.md (auto-loaded). Scan index for relevant memory files based on session topic. Read up to 3 relevant files.
2. Git state -- Branch, status, recent commits, diff against main.
3. Plan state -- Check your project's docs/plans/ for active plans and any session tracking document.
4. Skill loading -- Domain-classify the work, load matching skills per your project's domain-to-skill map.
5. SO2 pre-session -- Check GitHub issues, PR status, open blockers.

Session Boundary Rules

Token cap (configurable) enforces session budget:

• Yellow (~70% capacity): Finish current task, plan exit
• Red (~85% capacity): No new chunks, commit what's done
• Critical (~95% capacity): Commit NOW, tell user to resume

At EVERY phase transition, check context budget. If in red/critical zone:

"Session cap approaching. Phase N complete. Resume: /SO-godspeed-resume P{N+1}"

Memory Protocol

See references/memory-protocol.md for the full protocol.

On P0 (load): Read MEMORY.md + relevant topic files. During execution: Save feedback immediately, save decisions with absolute dates. On P8 (close): Review session for memorable patterns, update or create memory files.

What This Replaces