Add /claude-code-for-teams, the rollout page for the team question - #156
Conversation
Targets the claude code for teams cluster: the searcher is a lead or an admin who already has Claude Code working for one person and wants to know what changes for ten. Anthropic answers that across about a dozen pages, and this one collects them in the order a rollout hits them. Seven tables: the per-plan admin matrix, the five settings layers in precedence order, the CLAUDE.md scopes, the skill locations, the plugin and marketplace mechanisms, the MCP scopes, and Anthropic's own table of what a repository supplies before the folder itself is trusted. Then a seven step rollout with a copyable team settings file, the selection half that no vendor mechanism covers, and six declared gaps. Differentiated from its neighbours on purpose. /manage-ai-skills keeps the cross-vendor organisation view and this page drills into the Claude Code column; /claude-skills keeps the format; /agent-skills-support keeps the client matrix; /guides/how-to-write-a-skill-md keeps the authoring rules and this page links to it instead of repeating them. Every claim was read on 2026-08-24 from code.claude.com and the Anthropic help center. Four honest findings are in the page rather than smoothed over: the enterprise skills location has no published path, the skill precedence runs the opposite way from the settings precedence, the plan matrix has no column for seat type, and Anthropic states outright that Claude Code has no built-in registry to browse.
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
Important
This repository does not receive automatic reviews because it has fewer than 10 stars. ⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (15)
Included review availability: Your plan provides up to 1 included review per hour; 0 remain after this review. WalkthroughAdds a Claude Code for Teams resource page with typed content, article rendering, SEO metadata, social image routes, Markdown access, resource registration, related links, CTA analytics locations, and comprehensive tests. ChangesClaude Code for Teams resource
Estimated code review effort: 4 (Complex) | ~45 minutes Merge Risk: ⚪ Minimal · up to This PR adds a documented rollout page and supporting SEO and navigation changes, with the supplied checks passing. No actionable merge-blocking risk remains beyond normal review. Sequence Diagram(s)sequenceDiagram
participant Browser
participant NextRoute
participant ResourceShell
participant ClaudeCodeForTeamsPage
participant MarkdownAPI
Browser->>NextRoute: request /claude-code-for-teams
NextRoute->>ResourceShell: apply page layout
ResourceShell->>ClaudeCodeForTeamsPage: render typed resource content
ClaudeCodeForTeamsPage-->>Browser: return HTML and SEO metadata
Browser->>NextRoute: request Markdown with Accept header
NextRoute->>MarkdownAPI: rewrite request with resource path
MarkdownAPI-->>Browser: return Markdown content
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
|
@coderabbitai full review |
|
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 1f3679c53f
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| label: "Install the plugin once per team", | ||
| body: "Run /plugin marketplace add TommyBez/skillsboard and then /plugin install skills-board@skills-board, and Claude Code has the team library in session. A team that wants it on every clone can put the same marketplace in extraKnownMarketplaces and the plugin in enabledPlugins in the committed .claude/settings.json, which is exactly the mechanism the plugins section above describes.", |
There was a problem hiding this comment.
For a fresh teammate cloning the repository, this is not an installation performed “once per team”: the page's own plugins section explains that an external plugin enabled only through project settings remains uninstalled and prompts that teammate to run the install command. Committing extraKnownMarketplaces and enabledPlugins therefore does not put the team library into every clone; revise this guidance to require a per-user install or provisioning through the seed directory.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Founded, fixed in b8ea9c8. The path copy contradicted the plugins section of the same page, which already documents the v2.1.195 behavior: an external plugin enabled only through committed project settings does not load until each teammate runs the printed install command. The label no longer says once per team, and the body now states the install is per user, that a committed extraKnownMarketplaces plus enabledPlugins produces a per teammate prompt with the exact command on first run, and that the seed directory is the promptless provisioning route for containers and CI. Tests 495/495 and typecheck clean on the new head.
|
@CodeRabbit review |
✅ Action performedReview finished.
|
The Skills Board path section claimed committing extraKnownMarketplaces and enabledPlugins installs the plugin once for the whole team. The plugins section of the same page documents the opposite for external sources since v2.1.195: each teammate gets an install prompt on first run. Align the path copy with that behavior and with the seed directory provisioning route. Raised by Codex review.
1 participant
Hypothesis
A lead or an admin searching "claude code for teams" already has Claude Code working for one person. They want to know what changes for ten: which plan carries the admin features, where configuration lives, what a teammate gets from a clone, and what they have to install by hand. Anthropic answers all of that, across about a dozen pages, and almost none of it is on the page that ranks for the phrase.
This page collects those answers in the order a rollout hits them, and it is the natural place for Skills Board to appear: the vendor mechanisms solve distribution, and the documentation says outright that the catalog of what is approved is left to each organisation.
Target from the phase 2 keyword pack: "claude code for teams" (210/month, KD 21), with "claude code enterprise" (880, KD 35), "skill library" (480, KD 0) and "skill sharing" (110, KD 0) as secondaries. Raw sum 1,680, realistic 300 to 400.
What is on the page
Seven tables, each one a question a rollout has to answer:
claudeMdkey,claudeMdExcludes, and the enforcement versus guidance line Anthropic draws itself.--add-dir, and claude.ai sync.extraKnownMarketplaces,enabledPlugins, the plugin seed directory, and the four managed lockdown keys.Then a seven step rollout with a copyable team
.claude/settings.json, the selection half that has no setting, six declared gaps, and seven FAQ entries.Honest findings, in the page rather than smoothed over
managed-settings.json,managed-settings.d/andmanaged-mcp.jsonand never names a skills folder; the settings reference calls it "the managed policy directory" without saying where that is.deployskill beats the one the team committed. Settings: project local and shared project both sit above user. Both are documented clearly and no page reconciles them.strictPluginOnlyCustomizationwithskillsin the list stops loading.claude/skills, which is the exact opposite of the commit-it-to-the-repository advice on the skills page. Both are published as good practice.docs.claude.com/en/docs/claude-code/*now redirects tocode.claude.com/docs/en/*. The page cites the destination.Sixteen first-party sources, all read on 2026-08-24: fifteen code.claude.com pages and one Anthropic help center article dated June 11, 2026. Every source is cited by at least one section and every section cites at least one source, both enforced by tests.
Differentiation
/manage-ai-skillskeeps the cross-vendor organisation view; this page drills into the Claude Code column./claude-skillskeeps the format; this page keeps the rollout./agent-skills-supportkeeps the client matrix./guides/how-to-write-a-skill-mdkeeps the authoring rules and this page links to it rather than repeating them.Two contextual inbound links added, from
/claude-skillsand/manage-ai-skills. No link from the homepage.Checks
node --test tests/*.test.mjs).tsc --noEmitexit 0, no new errors.next build), plus anext startsmoke test:/claude-code-for-teams200, twin.md200text/markdown,Accept: text/markdownnegotiation 200, trailing slash 308 to the canonical, canonical andrel=alternatein the head, TechArticle and FAQPage in the JSON-LD, entry present in/sitemap.xml,/llms.txtand the/resourceshub, opengraph image 200.Deviations, declared
/best-claude-skills7,598,/vercel-skills7,130,/copilot-skills8,792). The excess is in table cells rather than prose.Last reviewedline inllms.txtstays at 2026-08-19. This page re-read only its own sources, so moving it would assert a full re-review that did not happen. Same reasoning and same test as/agent-skills-support,/manage-ai-skills,/vercel-skillsand/copilot-skills.-skills, so it needs its ownAccept: text/markdownrewrite innext.config.ts, the way/skill-examplesdoes. Covered by a test.Metric and how we will know
Metric: GSC impressions and clicks on the "claude code for teams" query family (claude code for teams, claude code team plan, claude code enterprise, claude code managed settings, claude code for teams setup).
Baseline today: zero. The family does not appear in any GSC query list we have pulled; the 21 to 24/08 window shows 63 distinct queries and 278 impressions with none of them in this cluster, and the site's dominant query is still "manage ai skills across organisation".
In 2 to 4 weeks we expect:
/manage-ai-skills, was written for a family that already had 208 impressions; this one is written for a family that has none, so impressions appearing at all is the finding.Kill criteria, stated in advance: if the URL is indexed and the family is still at zero impressions after 4 weeks, the head term is being taken by Anthropic's own documentation and the answer is not another page on the same intent.
Summary by CodeRabbit
New Features
Tests