name	onboarding-guide
description	Guide new developers through team onboarding. Analyzes project health, team patterns, delivery metrics, and codebase structure to produce a personalized onboarding plan with specific first tasks, graduated complexity, and human connections. A guided walkthrough, not an agent identity. Use when: "onboard a new developer", "getting-started guide", "team orientation for new hire", "first week plan", "day-one productivity".
phase_relevance	["build","review","operate"]
archetype_relevance	["*"]

Onboarding Guide

Produces a personalized onboarding path for a new developer by combining team health signals, codebase structure, and sprint state into a structured week-by-week plan with concrete first tasks.

Quick Start

Invoke this skill when someone new is joining the team. Typical outputs:

Team health snapshot (sprint state, velocity, blockers)
Recommended first tasks (ranked by learning value + risk + impact)
Key systems overview (what to learn in what order)
Team patterns (branching, review, testing, communication)
Structured weekly plan (orientation → first contribution → independence)
People to connect with (who owns what)

First Strategy: Use wicked-* Ecosystem

Delivery reports: /wicked-garden:delivery:report for team health overview
Memory: wicked-brain:memory (recall mode) for team conventions and past decisions
Search: wicked-garden:search for codebase navigation (backed by the codebase-narrator skill for architectural overview)
Tasks: Inspect native tasks via TaskCreate/TaskUpdate with metadata={event_type, chain_id, source_agent, phase}

Process

1. Gather Team Context

From delivery reports (if CSV/export available):

/wicked-garden:delivery:report {data_file}

From native tasks: Read session tasks under ${CLAUDE_CONFIG_DIR}/tasks/{session_id}/ — filter by metadata.event_type=="task" for human-visible items.

From memory:

Skill(skill="wicked-brain:memory", args="recall \"team conventions\"")
Skill(skill="wicked-brain:memory", args="recall \"architecture decisions\"")
Skill(skill="wicked-brain:memory", args="recall \"onboarding\"")

From codebase (FTS5 over indexed code, docs, wiki):

wicked-brain:search README
wicked-brain:search "TODO|FIXME|HACK"

2. Assess Team Health

Sprint status: in flight, blocked, done
Velocity: tasks/week, points/sprint
Blockers: what's stuck and why
Team capacity: who's working on what

Present as a concise snapshot, not a data dump.

3. Identify Good First Tasks

Rank candidates by:

Learning value — introduces key systems
Risk — low-risk preferred (isolated blast radius)
Team impact — helpful contributions preferred
Independence — minimal coordination needed

Sources: unassigned low-priority items, documentation gaps, small bugs, test coverage gaps.

4. Map Key Systems

Create a systems overview:

Core services — what the project does
Architecture — high-level component map
Data flow — how data moves through the system
Deployment — how code ships to production
Dependencies — external services and integrations

Hand off to the codebase-narrator skill for the full architectural walk.

5. Explain Team Patterns

Document:

Branching strategy (trunk-based, git-flow, etc.)
Review process (PR template, required reviewers, merge criteria)
Testing approach (what tests are expected per change)
Communication (where decisions happen — Slack, meetings, docs)
On-call / support (how incidents are handled)

6. Build the Week-by-Week Plan

See refs/plan-template.md for the full structure.

Shape:

Week 1 — Orientation: read docs, set up env, meet people, pick first task
Week 2 — First Contribution: complete first task with pair programming, submit first PR, attend sprint ceremonies
Week 3 — Independence: take a medium-complexity task, review someone else's PR, identify an improvement, share feedback

Quality Standards

Good onboarding plans have:

Concrete first tasks — not "explore the codebase" but specific tickets
Context before code — team health and patterns before diving in
Graduated complexity — easy → medium → harder
Human connections — who to talk to about what
Feedback loop — check-in points to adjust the plan

Common Pitfalls

Information overload on day one
Assuming knowledge of team-specific tools
Skipping team culture and patterns
Starting with the hardest problem
No check-in points to course-correct