codingthefuturewithai/teamcraft-jcg:onboard

Goal

Read the team member's environment and present it through the TeamCraft lens. Show them what exists, describe the current state, and explain how TeamCraft applies to their role going forward. This is orientation — not a handoff to work.

Hard Constraints

• Read-only. Advisory only. This skill observes and explains. It never offers to create, update, or run any workflow. It never says "want me to help you work on that?" The session ends at orientation.
• Atlassian MCP and GitHub connections are verified through real tool calls, not by asking.
• This is Cowork compatible — no local filesystem tools are used.
• The moment a sentence begins with "shall I" or "want me to" — stop. That is out of scope.

Establish Role

Ask the team member their role. Everything that follows is calibrated to it.

• Developer — build loop, issue work, code, PRs
• Product Manager — requirements, sprint planning, visibility
• Tech Lead / Architect — decisions, conventions, code review
• QA Analyst — sprint awareness, test prep, bug capture
• DevOps Engineer — CI/CD setup and health

If they don't have a role or want the full picture, give the overview for all roles.

Verify Connections

Attempt real calls. Do not ask whether things are configured.

Atlassian MCP (Jira + Confluence): Use mcp__sooperset-mcp-atlassian__jira_get_all_projects. If it returns results, the server is up and authenticated. If not, explain what needs to be configured — in plain terms appropriate to the role.

GitHub: Try in this order:

1. Run gh auth status via Bash. If it succeeds, the CLI is authenticated — use gh throughout.
2. If Bash is unavailable or gh is not found, look for a GitHub MCP connector in your available tools and attempt to list repos or verify access through it.
3. If neither works, note GitHub access is unavailable and proceed with Jira and Confluence data only. Do not block orientation on GitHub — surface what's available and note the gap.

If a connection fails, guide calmly through what is needed without assuming technical familiarity.

Read the Environment

With connections confirmed, read what exists. The goal is breadth, not depth — understand the shape of the environment, not every detail.

Jira:

• Surface what projects are visible. Ask the team member which one (or ones) they are joining.
• For the relevant project(s): use mcp__sooperset-mcp-atlassian__jira_get_agile_boards to find the project's board, then mcp__sooperset-mcp-atlassian__jira_get_sprints_from_board to list sprints. Surface the active sprint and ask the user to confirm. If a sprint is confirmed, use mcp__sooperset-mcp-atlassian__jira_get_sprint_issues to get the sprint's issue list. If no sprint is identified, use mcp__sooperset-mcp-atlassian__jira_search with JQL project = [PROJ] AND resolution = Unresolved to describe the overall state.

GitHub:

• If gh CLI is available: run gh repo list via Bash to surface accessible repos. Ask which one they are joining if ambiguous.
• If using a GitHub MCP connector: use it to list or search for repos accessible to the user.
• If GitHub is unavailable: note this in the presentation — the developer will need to verify their GitHub connection separately before using build loop skills.

Confluence:

• Use mcp__sooperset-mcp-atlassian__confluence_search broadly to find the project's key artifacts: PRD, tech decisions, conventions. If found, use mcp__sooperset-mcp-atlassian__confluence_get_page with the page ID from the search results to read enough to characterise their state — what phase they're in, how complete they appear, what they cover. Not a full summary — just enough to speak to their existence and maturity.
• If artifacts are missing or incomplete, note that clearly — it is relevant context for the orientation.

Present What You Found

Give the team member a clear picture of what you can see:

• What Jira projects exist and which one they're joining
• The current state of the sprint and open issues (as much as was discoverable)
• What GitHub repos exist and which one is relevant
• What Confluence artifacts exist: which documents are there, what state they appear to be in, what is missing

Be direct. "You have a Jira project with 8 open issues across the backlog and a PRD that covers the compression service scope. Tech decisions and conventions documents don't exist yet." That is more useful than a long qualified description.

Advise Through the TeamCraft Lens

Now map what you found onto TeamCraft — for their role specifically.

The question to answer: given what exists in this environment right now, how does TeamCraft apply and where does this person fit?

• If there are open issues and they're a developer: "The build loop is your entry point. Issues are waiting. fetch-issue is how you begin."
• If there's a PRD but no sprint and they're a PM: "The PRD exists. The next step in the TeamCraft upstream flow is tech decisions — but if those are captured, plan-sprint is what turns the PRD into a Jira backlog ready for sprint assignment."
• If artifacts are missing and they're a tech lead: "Tech decisions and conventions aren't documented yet. Those are your first two artifacts in TeamCraft — and they flow directly into every developer's planning session."
• If it's early and little exists: describe what TeamCraft would build from here and in what order.

Explain the map. Don't offer to navigate it.

Close

End with one sentence naming where this person would begin when they're ready to work — the specific skill, nothing more.

"When you're ready to start, teamcraft-jcg:fetch-issue is your entry point." Full stop. No offer to run it now.