Needle

@needle list all my collections and show me which ones have the most documents indexed

Based on your Needle account, you have 4 active collections. 'Product Documentation' contains 127 indexed documents (2.3 GB), last updated 3 days ago. 'Customer Support Tickets' has 89 documents (1.1 GB), updated yesterday. 'Marketing Assets' contains 34 documents (890 MB), and 'Internal Wiki' has 12 documents (340 MB). The Product Documentation collection is your largest by both document count and storage size.

This example demonstrates read-only discovery of your Needle workspace structure. Requires a valid API key with read permissions. Useful for auditing what content you've already indexed before creating new collections or uploading files. Response times depend on the number of collections in your account.

@needle create a new collection called 'Q1 2025 Planning' for our quarterly strategy documents

I've created the collection 'Q1 2025 Planning' in your Needle workspace (collection ID: coll_8x9k2m4n). The collection is now ready to accept document uploads. You can add files through the Needle dashboard or via API. Once documents are uploaded, they'll be indexed and searchable within your Needle environment, typically within a few minutes depending on file size and count.

This write operation creates a new organizational container in Needle. The MCP confirms the collection name before executing to prevent accidental creation. You'll need API key permissions that allow collection creation. After creation, you must upload files separately—this action only establishes the collection structure.

@needle show me the stats for my 'Customer Support Tickets' collection and summarize what kind of content volume we're dealing with

Your 'Customer Support Tickets' collection contains 89 indexed documents totaling 1.1 GB of storage. The collection was created 6 months ago and last updated yesterday at 4:32 PM UTC. Index size is 340 MB. Given this volume—roughly 15 tickets per month—you're maintaining a moderate support knowledge base. The recent update suggests active use, and the 3:1 compression ratio (1.1 GB → 340 MB index) indicates text-heavy documents rather than large attachments.

This example pairs Needle's metadata retrieval with AI analysis to contextualize raw statistics. The stats endpoint returns timestamps, document counts, and storage metrics—the AI layer interprets trends and scale. Requires the collection ID, which you can get from a list operation first. Stats update in near-real-time as you add or remove files.

When Needle makes sense for support ticket deflection

A 6-person support team fields 200 tickets a week, half of which repeat the same product questions. They upload their help docs and internal runbooks to a Needle collection, then wire it into their Switchy workspace so agents can query "how do we handle refund requests over $500" mid-ticket without leaving the thread. The 5 tools cover the basics: create a collection per product line, list files to audit what's indexed, delete outdated PDFs when policies change. The API key auth means no OAuth dance, so setup takes 10 minutes. This works if your knowledge base is under 10,000 documents and you're not doing real-time updates every hour. Beyond that threshold, you'll want a dedicated vector store with webhook sync. If your team closes 50+ tickets a day and half involve doc lookup, Needle pays for itself in saved tab-switching.

Needle as a lightweight onboarding assistant

A 12-person startup hires 3 engineers in a month and realizes their onboarding Notion is a maze of dead links. They create a Needle collection called "New Hire Docs", upload the 40-page handbook and 15 how-to videos, then share a Switchy thread where new hires ask "where's the staging deploy guide" and get instant answers. The collection stats tool shows them which docs get queried most, so they know what to expand. This setup shines when onboarding happens in batches (quarterly cohorts, not rolling weekly hires) and your docs change monthly, not daily. If you're onboarding 10+ people a month or your handbook lives in 5 different tools, Needle's single-collection model gets clunky fast. For small teams with stable docs, it's a 2-hour fix that cuts onboarding Slack noise by half.

When Needle beats a shared Google Drive for sales intel

A 4-person sales team competes against 8 vendors and keeps losing deals because reps can't remember which competitor lacks SOC 2 compliance or which one just raised a Series B. They dump 30 competitor PDFs and earnings call transcripts into a Needle collection, then query it during discovery calls: "what's Competitor X's pricing model for enterprise?" The list-files tool lets the sales lead audit what's indexed before a big pitch. This works when your intel updates monthly (new funding rounds, product launches) but not hourly (live pricing changes). If your team runs 20+ discovery calls a week and competitive intel is the difference between winning and losing, Needle turns tribal knowledge into a searchable asset. Beyond 100 competitor docs or real-time pricing wars, you need a CRM integration instead.