Search

Search Command

If you see unfamiliar placeholders or need to check which tools are connected, see CONNECTORS.md (opens in a new tab).

Search across all connected MCP sources in a single query. Decompose the user’s question, run parallel searches, and synthesize results.

Instructions

1. Check Available Sources

Before searching, determine which MCP sources are available. Attempt to identify connected tools from the available tool list. Common sources:

• ~~chat — chat platform tools
• ~~email — email tools
• ~~cloud storage — cloud storage tools
• ~~project tracker — project tracking tools
• ~~CRM — CRM tools
• ~~knowledge base — knowledge base tools

If no MCP sources are connected:

To search across your tools, you'll need to connect at least one source.
Check your MCP settings to add ~~chat, ~~email, ~~cloud storage, or other tools.

Supported sources: ~~chat, ~~email, ~~cloud storage, ~~project tracker, ~~CRM, ~~knowledge base,
and any other MCP-connected service.

2. Parse the User’s Query

Analyze the search query to understand:

• Intent: What is the user looking for? (a decision, a document, a person, a status update, a conversation)
• Entities: People, projects, teams, tools mentioned
• Time constraints: Recency signals (“this week”, “last month”, specific dates)
• Source hints: References to specific tools (“in ~~chat”, “that email”, “the doc”)
• Filters: Extract explicit filters from the query:
  • from: — Filter by sender/author
  • in: — Filter by channel, folder, or location
  • after: — Only results after this date
  • before: — Only results before this date
  • type: — Filter by content type (message, email, doc, thread, file)

3. Decompose into Sub-Queries

For each available source, create a targeted sub-query using that source’s native search syntax:

~~chat:

• Use available search and read tools for your chat platform
• Translate filters: from: maps to sender, in: maps to channel/room, dates map to time range filters
• Use natural language queries for semantic search when appropriate
• Use keyword queries for exact matches

~~email:

• Use available email search tools
• Translate filters: from: maps to sender, dates map to time range filters
• Map type: to attachment filters or subject-line searches as appropriate

~~cloud storage:

• Use available file search tools
• Translate to file query syntax: name contains, full text contains, modified date, file type
• Consider both file names and content

~~project tracker:

• Use available task search or typeahead tools
• Map to task text search, assignee filters, date filters, project filters

~~CRM:

• Use available CRM query tools
• Search across Account, Contact, Opportunity, and other relevant objects

~~knowledge base:

• Use semantic search for conceptual questions
• Use keyword search for exact matches

4. Execute Searches in Parallel

Run all sub-queries simultaneously across available sources. Do not wait for one source before searching another.

For each source:

• Execute the translated query
• Capture results with metadata (timestamps, authors, links, source type)
• Note any sources that fail or return errors — do not let one failure block others

5. Rank and Deduplicate Results

Deduplication:

• Identify the same information appearing across sources (e.g., a decision discussed in ~~chat AND confirmed via email)
• Group related results together rather than showing duplicates
• Prefer the most authoritative or complete version

Ranking factors:

• Relevance: How well does the result match the query intent?
• Freshness: More recent results rank higher for status/decision queries
• Authority: Official docs > wiki > chat messages for factual questions; conversations > docs for “what did we discuss” queries
• Completeness: Results with more context rank higher

6. Present Unified Results

Format the response as a synthesized answer, not a raw list of results:

For factual/decision queries:

[Direct answer to the question]

Sources:
- [Source 1: brief description] (~~chat, #channel, date)
- [Source 2: brief description] (~~email, from person, date)
- [Source 3: brief description] (~~cloud storage, doc name, last modified)

For exploratory queries (“what do we know about X”):

[Synthesized summary combining information from all sources]

Found across:
- ~~chat: X relevant messages in Y channels
- ~~email: X relevant threads
- ~~cloud storage: X related documents
- [Other sources as applicable]

Key sources:
- [Most important source with link/reference]
- [Second most important source]

For “find” queries (looking for a specific thing):

[The thing they're looking for, with direct reference]

Also found:
- [Related items from other sources]

7. Handle Edge Cases

Ambiguous queries: If the query could mean multiple things, ask one clarifying question before searching:

"API redesign" could refer to a few things. Are you looking for:
1. The REST API v2 redesign (Project Aurora)
2. The internal SDK API changes
3. Something else?

No results:

I couldn't find anything matching "[query]" across [list of sources searched].

Try:
- Broader terms (e.g., "database" instead of "PostgreSQL migration")
- Different time range (currently searching [time range])
- Checking if the relevant source is connected (currently searching: [sources])

Partial results (some sources failed):

[Results from successful sources]

Note: I couldn't reach [failed source(s)] during this search.
Results above are from [successful sources] only.

Notes

• Always search multiple sources in parallel — never sequentially
• Synthesize results into answers, do not just list raw search results
• Include source attribution so users can dig deeper
• Respect the user’s filter syntax and apply it appropriately per source
• When a query mentions a specific person, search for their messages/docs/mentions across all sources
• For time-sensitive queries, prioritize recency in ranking
• If only one source is connected, still provide useful results from that source