Enterprise Search (One Query Across Every Tool)
One query across every connected tool — email, chat, docs, wikis — decomposed, searched in parallel, synthesized with source attribution.
Before & after
Same prompt, same agent — the only difference is the skill.
Let me help you find that information. You'd need to search through Slack for any API design discussions, then check your email for related threads, scan any shared documents or wikis, and check your project tracker for related tickets. I'd recommend starting with the most recent conversations and working backward. You might find the discussion spread across multiple channels and platforms. Would it help if I suggested a search strategy, or would you prefer I help you pull from one source at a time? Once you've gathered the information, I can help synthesize it for you.
SEARCH DECOMPOSED INTO: → Design decision thread → Email follow-ups with specs → Updated API documentation → Implementation tickets RESULTS WITH SOURCE ATTRIBUTION: DECISION (engineering channel, Jul 18 3:14pm): Team chose REST over GraphQL. Rationale: simpler onboarding, lower infrastructure cost for current scale. FOLLOW-UP SPEC (email from Sarah, Jul 19 10:22am): Rate limiting set at 1000 requests per minute per key. Versioning via URL path: /v1/endpoints, /v2/endpoints. DOCUMENTATION UPDATE (API design doc, last edited Jul 22): Full specification reflects REST decision. Includes authentication flow, error codes, deprecation policy. IMPLEMENTATION TICKETS (project tracker): • Migrate existing GraphQL resolvers to REST handlers (assigned to 3 engineers) • Update client libraries for v2 endpoints (in progress) • Deprecate GraphQL API with 6-month window (planned) KEY DETAILS FOUND: Endpoint count: 47 total (23 migrated, 24 pending) Estimated completion: August 15 Breaking changes: authentication schema revised, error response format standardized
About this skill
name: kw-enterprise-search description: Use when one query across every connected tool — email, chat, docs, wikis — decomposed, searched in parallel, synthesized with source attribution.
Enterprise Search (One Query Across Every Tool)
Anthropic's Enterprise Search plugin searches all connected tools at once. Claude decomposes your question, runs targeted searches across every source, and synthesizes results into a single coherent answer with source attribution. Example: ask 'what did we decide about the API redesign' and Claude pulls the #engineering thread from Tuesday, the follow-up email from Sarah, and the updated API design doc, then summarizes: the team decided on Tuesday to go with REST over GraphQL, Sarah sent the spec Thursday, the design doc reflects the final approach. Connectors: Slack, Notion, Guru, Jira, Asana, Microsoft 365.
What you get
- Question decomposition into parallel searches
- simultaneous search across chat, email, docs, wikis
- synthesized single answer with source attribution
- per-source citations (chat thread, email, doc)
- connectors Slack, Notion, Guru, Jira, Asana, Microsoft 365
Customize your output
- Connectors: Slack, Notion, Guru, Jira, Asana, Microsoft 365
- Query scope: decision, doc, spec, discussion
- Output: single synthesized answer with citations
Example output
'What's the current PTO policy?' returns the current HR handbook section, the Slack thread where the leadership team clarified an edge case, and a Jira ticket adjusting the accrual formula — all cited.
Best for
Anyone at a company drowning in tool sprawl who needs to answer 'where did we decide X' fast
SKILL.md preview
---
name: kw-enterprise-search
description: Use when a question needs an answer synthesized from multiple connected tools at once, such as email, chat, docs, and wikis, rather than searching one source at a time.
version: 1.0.0
category: Search / Knowledge
author: AgentVolt
license: proprietary
tags:
- search
- knowledge
- p3
---
# Enterprise Search (One Query Across Every Tool)
Decomposes a question, runs targeted searches across every connected source, and synthesizes results into one coherent answer with source attribution.
## When to use
… (sign up to view the full skill)