Personas — Smart SDLC

Personas are named AI roles, each with a defined skill set, communication style, and domain focus. Every team member picks one persona as their primary_persona during npx @supermldev/smart-sdlc persona.

The 6 Personas

Scout — Code Archaeologist

Scout is the universal starting point for any brownfield project. It reads the existing codebase and produces living documentation — structure maps, architecture summaries, API references, and reverse-engineered ADRs.

Primary skills:

When to use: Always at the start of any brownfield project, before any other persona activates. Also useful when onboarding new team members.

Product / BA — Agent PM

The Product Manager / Business Analyst persona handles requirements discovery, product thinking, and documentation. It bridges business needs and technical execution.

Primary skills:

Cross-persona collaboration:

• Call @sml-agent-architect when architecture and technical design decisions are needed
• Call @sml-agent-developer to validate implementation feasibility

Architect

The Architect persona translates requirements into technical designs. It creates architecture documents, Architecture Decision Records (ADRs), and breaks work into implementable epics and stories.

Primary skills:

Prerequisites: Architect expects a completed PRD (prd_complete: true in config.yml). If no PRD exists, activate @sml-agent-pm first.

Cross-persona collaboration:

• Call @sml-agent-pm for requirements clarification and product decisions
• Call @sml-agent-developer to validate implementation approach

Developer

The Developer persona implements stories using test-driven development, reviews code, and manages technical debt. It operates sprint by sprint.

Primary skills:

Prerequisites: Developer expects epics and stories to exist (epics_complete: true). Use @sml-agent-scout first if working on an unfamiliar codebase.

Cross-persona collaboration:

• Call @sml-agent-architect for design decisions during implementation
• Call @sml-agent-pm to clarify acceptance criteria

Modernization Lead — Agent Sage

The Modernization Lead handles legacy system analysis and migration planning. It has its own multi-phase workflow distinct from the standard SDLC — see Modernization Workflow for the full journey.

Primary skills:

Cross-persona collaboration:

• Call @sml-agent-architect for target architecture design decisions
• Call @sml-agent-pm for domain business rules and stakeholder requirements

Team Lead / PM — Agent Lead

The Team Lead / PM persona focuses on delivery coordination — breaking architecture into epics, planning sprints, managing capacity, and tracking progress.

Primary skills:

Cross-persona collaboration:

• Call @sml-agent-pm for requirements and product direction
• Call @sml-agent-architect for technical planning and feasibility

QA / Test Engineer

The QA persona validates that built outputs satisfy requirements, acceptance criteria, and quality standards. It operates after implementation and before release.

Primary skills:

Can read: stories, acceptance criteria, implementation outputs, architecture constraints

Cannot: invent requirements, silently fill requirement gaps, approve ambiguous stories without flagging them

Output artifacts: {output_path}/qa/test-plan.md, test-execution.md, bug-triage.md, qa-signoff.md

Prerequisite: implementation_signed_off: true in config.yml

Cross-persona collaboration:

• Call @sml-agent-developer to resolve bugs found during testing
• Call @sml-agent-pm to clarify ambiguous acceptance criteria

Release / Ops

The Release persona ensures the solution is deployable, observable, and operationally ready. It owns the deployment plan, runbook, and release sign-off. It operates after QA sign-off.

Primary skills:

Can read: architecture, implementation outputs, test summaries, environment requirements

Cannot: define business features, fill missing test coverage by assumption, approve unsafe release posture without flagging it

Output artifacts: {output_path}/release/release-checklist.md, deploy-runbook.md, release-notes.md

Prerequisite: qa_signed_off: true in config.yml

Cross-persona collaboration:

• Call @sml-agent-qa to confirm validation status before deployment
• Call @sml-agent-architect for deployment topology questions

Persona Lifecycle

Personas are installed, not casually switched. Each activation is intentional and traceable.

Installing a Persona

npx @supermldev/smart-sdlc persona

Runs the setup wizard — captures your name, role, AI tool preference, and skill level. Writes _superml/persona.yml. Every activation is logged to _superml/audit.log.

Checking Your Installed Persona

npx @supermldev/smart-sdlc persona status

Displays your currently installed persona (name, role, tools, level) without running the wizard.

Exiting a Persona

npx @supermldev/smart-sdlc persona exit

Shows the current persona, confirms, then removes _superml/persona.yml. The exit event is logged to _superml/audit.log. Run this before switching roles.

Re-installing a Persona

Run npx @supermldev/smart-sdlc persona again. If a persona is already installed you are asked: “Re-install persona with new answers?” Switching to a different role shows a confirmation prompt before overwriting.

Audit Trail

Every install, exit, sml init, and sml reenter is appended to _superml/audit.log:

2026-04-23T09:00:00.000Z | persona-install | persona:product | phase:planning | user:Aria | project:my-app
2026-04-23T14:30:00.000Z | persona-exit    | persona:product | phase:planning | user:Aria | project:my-app
2026-04-23T14:31:00.000Z | persona-install | persona:architect | phase:solutioning | user:Rex | project:my-app

The file is git-ignored and never leaves your machine.

Persona Guard — How Skill Access Is Enforced

Each generated skill file (.github/skills/<name>/SKILL.md) contains a Persona Guard block. When the skill is activated:

1. The AI reads _superml/persona.yml silently
2. If primary_persona matches the skill’s allowed list → proceeds normally
3. If not → stops and presents this message:

This skill is outside your role. This skill is designed for: [Role Name]. Your current persona does not typically use it.

To collaborate with the right persona: /sml-meeting — bring other personas into the conversation

To see your own skills: /sml-help — context-aware guidance for your role

Access Map

Escape Hatch — Meetings

When you need a skill outside your role, use /sml-meeting to bring the correct persona into a shared AI session. This keeps collaboration explicit and traceable rather than letting any persona silently drift into another’s domain.

Choosing Your Persona

You can change your persona at any time:

npx @supermldev/smart-sdlc persona