Library
Claude Code Documentation (June 2026) · 15 of 15
Claude Code Documentation (June 2026)
ai LOW

Team Adoption — The Champion Role

Claude Code Documentation (June 2026) Anthropic
adoption champion onboarding rollout objections

Key Principle

A champion is an engineer already using Claude Code who spreads it as a multiplier, not a help desk: "You are acting as a multiplier for your team, not a help desk" (p. 2). Adoption that depends on one person is fragile; the goal is shared habits that compound on their own — "When questions in the channel are being answered by people other than you, the role has done its job" (p. 2).

Why This Matters

Scaling Claude Code to a team mirrors scaling work across context windows: share durable, copyable artifacts — skills are plain Markdown, so adoption is instant — not bespoke prompting (chunk 017 theme). The champion is "the warm introduction rather than the onboarding program."

Good Examples

Three behaviors (p. 2):

  1. Share what you discover — post prompts/screenshots/wins where the team already reads. "Documentation tells people what is possible; your posts show them what is actually working in your environment."
  2. Be the person people ask — answer with the actual prompt you used, so it's directly reusable.
  3. Grow the circle — set up lightweight recurring habits so momentum continues without you.

/team-onboarding scans your recent sessions, commands, and MCP servers and emits a guide a new teammate pastes as their first message to replay your setup; run it in a project you've spent real time in and pin the result (~2 min) (Section: Team adoption).

Sample posts (verbatim, p. 5):

"Learned today that @-mentioning a directory works. I pointed it at @src/components/ and asked which components were missing tests, and it surfaced two I had forgotten about." "Plan mode is the reason I am comfortable using this on code that matters. Press Shift+Tab until you see 'plan'; it lays out exactly which files it intends to touch before changing anything."

Counterpoints

  • Do not improvise security answers. For "What about security and data handling?", refer to your administrator; org policy is already configured. Champions must NOT improvise this answer (FAQ #7, p. 5).
  • Share reusable techniques (they compound as they spread), not status updates (they don't) (Section: What and where to share).
  • Format that works: screenshot + single line of context, or brief before/after. Long write-ups get saved-for-later and forgotten; short posts with screenshots get copied and tried (p. 4).

Key Commands & Config

# Weekly time budget (p. 3)
Posting wins/prompts        ~15 min  (screenshot + 1-2 sentences, no write-ups)
Answering in a channel      ~20 min  (answer publicly once, link back on recurrence)
Weekly show-and-tell thread ~5 min   (you post the prompt, team supplies content)
Optional pairing            0-30 min (reserve for genuinely blocked colleagues)

# Thirty-day rollout playbook (signal = community sustains itself without you)
1. Create channel, pin Quickstart (/docs/en/quickstart), post 2-3 own examples
   with prompts.            Signal: colleagues react/reply; >=1 question asked.
2. Weekly show-and-tell thread; answer every question publicly; share one custom
   skill or CLAUDE.md snippet.  Signal: someone else posts their own example.
3. Offer 2-3 short pairing sessions; consolidate Q&A into a pinned FAQ.
                                Signal: repeat usage (same people returning).
4. Identify a second champion; summarize what is/isn't working to your lead.
                                Signal: questions answered by people other than you.

# Quick-reference techniques (the "why")
/init           # generates CLAUDE.md; then add conventions, test commands,
                #   do-not-modify directories  (See Memory.)
Shift+Tab       # plan mode: Claude describes intended changes for approval
                #   before executing -- nothing modified until approved
@file  @dir/    # @-mention a file/directory to scope and feed context
Stop hook       # desktop notification when a long task completes  (See Hooks.)
SKILL.md        # save in .claude/skills/<name>/ to create a shared /name skill

Rules of Thumb

  • Answer with a prompt, not an explanation — colleagues learn more running it against their own problem (Section: Answering well).
  • Point at the feature, not the docs: "Try plan mode, press Shift+Tab until you see it" beats a doc link in the moment.
  • Objections: acknowledge → brief reframe → one concrete demo on the person's own code. "I tried it once and it hallucinated" → usually a context problem; re-run their prompt with @-mentioned files, /init, and actual error output (Scripted responses, chunk 017).
  • For a first task, recommend a real but contained, tedious-not-difficult chore being postponed (FAQ #1).

Related References