ai-dev-orchestration

---

LIBRARY-FIRST PROTOCOL (MANDATORY)

Before writing ANY code, you MUST check:

Step 1: Library Catalog

• Location: .claude/library/catalog.json
• If match >70%: REUSE or ADAPT

Step 2: Patterns Guide

• Location: .claude/docs/inventories/LIBRARY-PATTERNS-GUIDE.md
• If pattern exists: FOLLOW documented approach

Step 3: Existing Projects

• Location: D:\Projects\*
• If found: EXTRACT and adapt

Decision Matrix

Match	Action
Library >90%	REUSE directly
Library 70-90%	ADAPT minimally
Pattern exists	FOLLOW pattern
In project	EXTRACT
No match	BUILD (add to library after)

---

STANDARD OPERATING PROCEDURE

Purpose

Run full-stack AI-assisted development loops—framing, architecture, implementation, validation, and launch—while keeping confidence ceilings explicit and artifacts auditable.

Trigger Conditions

• Positive: end-to-end feature delivery, AI-assisted coding reviews, test orchestration, release-readiness drills, backlog triage for complex builds.
• Negative: single-file edits without coordination, pure prompt engineering (route to prompt-architect), or meta-skill creation (route to skill-forge).

Guardrails

• Skill-Forge structure-first: ensure SKILL.md, examples/, and tests/ exist; stage resources/ and references/ or log remediation tasks before completion.
• Prompt-Architect clarity: capture HARD/SOFT/INFERRED requirements, user acceptance criteria, and produce pure-English outputs with ceiling-aware confidence.
• SDLC safety: enforce branch/CI policy, code review checkpoints, test coverage gates, and rollback/feature-flag plans for risky changes.
• Adversarial validation: probe failure modes (flaky tests, dependency drift, latency regressions) and record evidence.
• MCP tagging: store orchestration logs under WHO=ai-dev-orchestration-{session} and WHY=skill-execution for reuse.

Execution Playbook

1. Intent & scope: map goals to release milestones; confirm non-functional targets (latency, reliability, compliance).
2. Plan & topology: assign planner, builder, reviewer, and tester agents; define lanes, SLAs, and review cadence.
3. Build & delegate: break work into increments, wire hooks to CI, and enforce registry-only agent usage.
4. Quality gates: run unit/integration/e2e suites, security/lint checks, and performance baselines with thresholds.
5. Adversarial loop: simulate rollback, partial failures, and migration paths; document deltas and evidence.
6. Delivery: summarize implemented changes, validation results, residual risk, and next actions with confidence ceiling.

Output Format

• Release objective, scope, and constraints.
• Role topology (planner/builder/reviewer/tester) with ownership and timelines.
• Delivery plan and checkpoints (CI hooks, review gates).
• Validation evidence (tests, performance, security) and risk log.
• Confidence: X.XX (ceiling: TYPE Y.YY) - rationale in pure English.

Validation Checklist

• Structure-first assets present or ticketed; examples/tests updated or planned.
• HARD/SOFT/INFERRED requirements addressed with owners and evidence.
• CI/review gates executed; rollback path documented; hooks within latency budgets.
• Adversarial and COV runs logged with MCP tags; confidence ceiling declared.

Completion Definition

Feature is deliverable when code, tests, and deployment assets meet acceptance criteria, risks are owned, rollback is ready, and orchestration notes persist in MCP.

Confidence: 0.70 (ceiling: inference 0.70) - SOP reframed with skill-forge structure, prompt-architect constraint handling, and SDLC orchestration guardrails.