project-iris 0.0.1

package/src/iris_bundle/frameworks/iris-core/memory-bank/glossary.md

@@ -0,0 +1,197 @@
+# irismd Glossary
+
+Core terminology used throughout irismd and AI-DLC.
+
+---
+
+## AI-DLC Core Concepts
+
+### Intent
+
+A high-level statement of purpose that encapsulates what needs to be achieved - whether a business goal, feature, or technical outcome. Intents serve as the starting point for AI-driven decomposition into actionable tasks.
+
+### Unit
+
+A cohesive, self-contained work element derived from an Intent. Units are designed to deliver measurable value and are loosely coupled, enabling autonomous development and independent deployment. Analogous to Subdomains in DDD or Epics in Scrum.
+
+### Bolt
+
+The smallest iteration in AI-DLC, designed for rapid implementation of a Unit or a set of tasks within a Unit. Bolts emphasize intense focus and high-velocity delivery, with build-validation cycles measured in hours or days rather than weeks. Analogous to Sprints in Scrum.
+
+### User Story
+
+A well-defined contract that aligns human and AI understanding of what needs to be built. Stories articulate functional scope and include acceptance criteria.
+
+---
+
+## AI-DLC Phases
+
+### Inception Phase
+
+The phase focused on capturing Intents and translating them into Units for development. Uses "Mob Elaboration" ritual. Outputs include PRFAQs, User Stories, NFRs, Risk descriptions, and Suggested Bolts.
+
+### Construction Phase
+
+The phase encompassing iterative execution of tasks, transforming Units into tested, operations-ready Deployment Units. Progresses through Domain Design → Logical Design → Code → Test.
+
+### Operations Phase
+
+The phase centered on deployment, observability, and maintenance of systems. AI analyzes telemetry, detects patterns, and proposes actionable recommendations.
+
+---
+
+## AI-DLC Rituals
+
+### Mob Elaboration
+
+A collaborative requirements elaboration and decomposition ritual in the Inception Phase. AI proposes breakdown of Intent into Stories and Units; humans (the mob) refine and validate.
+
+### Mob Construction
+
+A collaborative development ritual in the Construction Phase. Teams collocate, exchange integration specifications, make decisions, and deliver bolts.
+
+---
+
+## AI-DLC Artifacts
+
+### Domain Design
+
+An artifact that models the core business logic of a Unit independently of infrastructure. Uses DDD principles: aggregates, value objects, entities, domain events, repositories, factories.
+
+### Logical Design
+
+An artifact that translates Domain Designs by extending them to meet non-functional requirements using appropriate architectural patterns (CQRS, Circuit Breakers, etc.).
+
+### Deployment Unit
+
+Operational artifacts encompassing packaged executable code (containers, serverless functions), configurations (Helm Charts), and infrastructure components (Terraform, CloudFormation) that are tested for functional acceptance, security, and NFRs.
+
+### PRFAQ
+
+Press Release / FAQ document that summarizes business intent, functionality, and expected benefits.
+
+### NFR (Non-Functional Requirement)
+
+Requirements covering Performance, Scalability, Security, Reliability, Maintainability, and Observability.
+
+---
+
+## Bolt Types
+
+### DDD Construction Bolt
+
+A bolt type using Domain-Driven Design principles for complex business logic. Stages: Domain Model → Technical Design → ADR Analysis (optional) → Implement → Test.
+
+### Simple Construction Bolt
+
+A lightweight bolt type for UI, integrations, and utilities. Stages: Plan → Implement → Test.
+
+### Spike Bolt
+
+A bolt type for research and exploration when facing unknowns.
+
+---
+
+## irismd Architecture
+
+### Agent
+
+A specialized AI persona with defined responsibilities, skills, and workflows. irismd has four agents: Master (orchestrator, router, and project initializer), Inception, Construction, Operations.
+
+### Skill
+
+A focused capability within an agent. Skills are invoked to perform specific tasks (e.g., creating intents, gathering requirements, starting bolts).
+
+### Memory Bank
+
+File-based storage system for all project artifacts. Maintains context across agent sessions and provides traceability.
+
+### Slash Command
+
+Commands invocable in agentic coding tools that activate agents (e.g., `/irismd-master-agent`, `/irismd-inception-agent`).
+
+### Template
+
+Markdown templates used to create consistent artifacts (requirements, stories, unit briefs, etc.).
+
+### Facilitation Guide
+
+Documents that guide conversations for establishing standards (tech-stack, coding standards, etc.).
+
+### Mandatory Output Rules
+
+Formatting rules embedded at the top of every skill file. These rules ensure consistent output across all agents. Key rules include: no ASCII tables, numbered list format for options, and standard status indicators.
+
+### Status Indicators
+
+Visual markers used consistently across all agent outputs to show progress:
+
+- `✅` - Completed/done
+- `⏳` - In progress/current
+- `[ ]` - Not started/pending
+- `🚫` - Blocked/requires decision
+- `⚠️` - Warning/needs attention
+
+### Suggested Next Step
+
+A pattern used at the end of every skill output that presents available actions and recommends the most logical next action. Format: `→ **option** - Description`
+
+### Human Validation Point
+
+A checkpoint in agent workflows where the AI pauses and explicitly requests human approval before proceeding. Used to enforce the "Human Oversight as Loss Function" principle.
+
+---
+
+## Standards
+
+### Tech Stack
+
+Standard defining technology choices: languages, frameworks, databases, ORMs, authentication, infrastructure, package managers.
+
+### Coding Standards
+
+Standard defining code quality: formatting, linting, naming conventions, file organization, testing strategy, error handling, logging.
+
+### System Architecture
+
+Standard defining high-level design: architecture style, API design, state management, caching, security patterns.
+
+### UX Guide
+
+Standard defining UI/UX patterns: design system, styling approach, accessibility, responsive strategy.
+
+### API Conventions
+
+Standard defining API design: API style, versioning, response format, error format, pagination.
+
+---
+
+## Agentic Coding Tool Integrations
+
+### Claude Code
+
+Anthropic's CLI tool for Claude. irismd installs commands to `.claude/commands/`.
+
+### Cursor
+
+AI-powered code editor with agentic capabilities. irismd installs commands to `.cursor/commands/`.
+
+### GitHub Copilot
+
+GitHub's AI pair programmer. irismd installs commands to `.github/copilot/commands/`.
+
+---
+
+## Key Principles
+
+### Human Oversight as Loss Function
+
+The principle that human validation at each step catches and corrects errors early before they snowball downstream.
+
+### AI Drives, Human Validates
+
+AI proposes plans, decompositions, and implementations; humans approve, validate, and confirm decisions.
+
+### Semantically Rich Context
+
+Each artifact builds context for the next step, ensuring AI has accurate information throughout the lifecycle.

package/src/iris_bundle/frameworks/iris-core/memory-bank/intents/001-multi-agent-orchestration/requirements.md

@@ -0,0 +1,129 @@
+---
+status: complete
+completed: 2025-12-25
+note: "Implemented before irismd dogfooding. Stories not tracked in AI-DLC."
+---
+
+# Intent: Multi-Agent Orchestration
+
+## Overview
+
+Implement a multi-agent system that orchestrates the AI-DLC workflow through specialized agents, each responsible for a specific phase of the development lifecycle.
+
+**Related Research**: See `research/approval-gates-simplification.md` for approval gates analysis and simplification proposal.
+
+---
+
+## Functional Requirements
+
+### FR-1: Agent System
+
+- FR-1.1: System SHALL provide four specialized agents: Master, Inception, Construction, Operations
+- FR-1.2: Each agent SHALL have a defined persona, responsibilities, and skills
+- FR-1.3: Agents SHALL be invocable via slash commands in agentic coding tools
+- FR-1.4: Agents SHALL read project context from Memory Bank on activation
+
+### FR-2: Master Agent
+
+- FR-2.1: Master Agent SHALL analyze project state from Memory Bank artifacts
+- FR-2.2: Master Agent SHALL determine current AI-DLC phase based on artifact state
+- FR-2.3: Master Agent SHALL route users to appropriate specialist agents
+- FR-2.4: Master Agent SHALL provide project health dashboard
+- FR-2.5: Master Agent SHALL facilitate project initialization via standards setup
+
+### FR-3: Inception Agent
+
+- FR-3.1: Inception Agent SHALL create and manage intents
+- FR-3.2: Inception Agent SHALL gather functional and non-functional requirements
+- FR-3.3: Inception Agent SHALL define system context and boundaries
+- FR-3.4: Inception Agent SHALL decompose intents into units using DDD principles
+- FR-3.5: Inception Agent SHALL create user stories with acceptance criteria
+- FR-3.6: Inception Agent SHALL plan bolts for execution
+- FR-3.7: Inception Agent SHALL validate outputs with human before proceeding
+
+### FR-4: Construction Agent
+
+- FR-4.1: Construction Agent SHALL execute bolts through defined stages
+- FR-4.2: Construction Agent SHALL load bolt type definitions (DDD, TDD, BDD, Spike)
+- FR-4.3: Construction Agent SHALL execute stages as defined in bolt type (never assume)
+- FR-4.4: Construction Agent SHALL validate stage completion before advancing
+- FR-4.5: Construction Agent SHALL load project standards during execution
+- FR-4.6: Construction Agent SHALL generate artifacts using templates
+
+### FR-5: Operations Agent
+
+- FR-5.1: Operations Agent SHALL verify construction completion before deployment
+- FR-5.2: Operations Agent SHALL build deployment artifacts
+- FR-5.3: Operations Agent SHALL deploy to environments (Dev → Staging → Prod)
+- FR-5.4: Operations Agent SHALL verify deployment success
+- FR-5.5: Operations Agent SHALL setup monitoring and observability
+
+### FR-6: Skill System
+
+- FR-6.1: Each agent SHALL have a defined set of skills
+- FR-6.2: Skills SHALL be defined in markdown files
+- FR-6.3: Skills SHALL be invocable by agent based on context
+- FR-6.4: Skills SHALL produce artifacts using templates
+
+---
+
+## Non-Functional Requirements
+
+### NFR-1: Performance
+
+- Agents SHALL activate within agentic coding tool response time limits
+- Skills SHALL execute without blocking user interaction
+
+### NFR-2: Usability
+
+- Agent personas SHALL be consistent and clear
+- Skill outputs SHALL be structured and predictable
+- Human validation points SHALL be clearly communicated
+
+### NFR-3: Maintainability
+
+- Agent definitions SHALL be in markdown format
+- Skill definitions SHALL be modular and independent
+- Template system SHALL allow customization
+
+### NFR-4: Extensibility
+
+- New agents SHALL be addable without modifying core system
+- New skills SHALL be addable to existing agents
+- New bolt types SHALL be definable via templates
+
+### NFR-5: Minimal Approval Gates
+
+- Human approval gates SHALL be minimized to essential decision points
+- Related artifacts SHALL be batched for single review where possible
+- Auto-validation SHALL be used when objective criteria exist (e.g., tests pass)
+- Progress display SHALL show workflow position at each gate
+- Agents SHALL be lightweight routers; skills SHALL contain complete behavior
+
+---
+
+## Acceptance Criteria
+
+### AC-1: Agent Invocation
+
+- GIVEN a user in an agentic coding tool with irismd installed
+- WHEN user invokes `/irismd-master-agent`
+- THEN Master Agent activates and analyzes project state
+
+### AC-2: Phase Routing
+
+- GIVEN a project with completed inception artifacts
+- WHEN Master Agent analyzes state
+- THEN Master Agent recommends Construction Agent with correct unit
+
+### AC-3: Skill Execution
+
+- GIVEN Inception Agent is active
+- WHEN user requests requirements gathering
+- THEN Agent executes requirements skill and produces artifacts
+
+### AC-4: Stage Validation
+
+- GIVEN Construction Agent executing a DDD bolt
+- WHEN domain design stage completes
+- THEN Agent validates completion criteria before advancing to logical design