How this skill is triggered — by the user, by Claude, or both
Slash command
/foundation:startThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Orient the user to Unstructured Foundation. This command is user-invoked only; do not use it for unsolicited capability messages.
Orient the user to Unstructured Foundation. This command is user-invoked only; do not use it for unsolicited capability messages.
First, call pipeline_processing_status with no since value.
Use the returned status to give the next step:
no_sources: Say no sources are connected yet. Suggest /foundation:connect, then processing files, then asking what is searchable.never_run: Say sources are connected but files have not been processed yet. Mention source names if the status includes them. If names are missing and pipeline_list_sources is available, call it.running: Say files are still being processed. Include progress counts if returned.failed: Say processing hit a problem. Summarize the status and do not retry automatically.ready: Say Foundation is ready. Offer short examples like "What documents do you have access to?", "Search my documents for quarterly revenue.", and "What's new since yesterday?"If sources are already connected, still mention that the user can add another source with /foundation:connect.
Do not call pipeline_link_source unless the user explicitly asks to connect a source now.
Use public language: say "Unstructured Foundation", "sources", "files", "documents", "processed", "searchable", and "data source". Avoid internal system terms in user-facing prose.
npx claudepluginhub unstructured-io/unstructured-foundation-marketplace --plugin foundationCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.