From first draft to published, tracked, and searchable — Commons covers the whole lifecycle of an AI asset, so your team ships from one trusted source and the person accountable for it always has an answer.
Every published skill, guardrail, template, and context doc — organized by team, function, and activity, with usage counts and version history built in. Semantic search means people find the right asset by describing what they need, not guessing at a filename.
Every submission is routed to the right reviewer automatically — a team approver for the asset's function, or an org admin as fallback. Claude generates a change summary, risk flags, and a recommendation before your reviewer even opens it. And reviewers don't have to live in another tab: submissions land in Slack with one-click Approve/Reject buttons.
See usage by team and asset type, catch published assets going stale before someone relies on outdated guidance, and get Pulse — a narrative summary written by Claude — summarizing what changed and what needs attention. Pulse findings now come with one-click fixes: archive a stale asset, draft an AI refresh, or merge duplicate tags — right from the report, always a human click to approve.
When several people keep needing something the library doesn't cover, Commons notices — and writes a complete first draft to fill the gap, not just an outline. Your team can also raise a hand directly: any suggestion from Ask lands in the same admin view. Either way admins get pinged the moment a gap surfaces, in Slack and the notification bell. It's built privacy-first: nobody's prompts are ever shown verbatim, only generalized themes aggregated across multiple people, and it's strictly double opt-in — an admin enables it for the org, then each person chooses to participate. Both are off by default.
Sometimes you don't want to search — you want an answer. Ask a question in plain language and Claude answers it grounded only in your published library, with citations back to the exact asset. Keep the thread going with follow-ups, and ask from anywhere: a launcher rides along on every page so nobody breaks their flow. When the library genuinely doesn't cover something, Ask doesn't guess — it turns the gap into action.
Every use is logged with an outcome, not just a count — did people actually apply the asset, or work around it? When enough people report friction, Claude drafts a revision and an admin approves it as a branch. And as the library grows, Commons watches for published assets that quietly contradict each other, flagging the pair before someone gets conflicting guidance.
Paste one connector URL and Claude can pull your published assets into any conversation — web, desktop, mobile, and Claude Code. Every use is logged automatically, so your analytics stay accurate without anyone lifting a finger, and Settings shows each person's live connection status.
Every change is routed, reviewed, and logged — so "what's live, who approved it, and when" is a lookup, not an investigation. The library flags its own gaps, conflicts, and friction, and pings you the moment something needs a call.
The governed library for your team's AI skills, prompts, and guardrails.