Search Context

Search Team Knowledge

Instructions

1. Parse the query

   • Extract the search query from $ARGUMENTS.
   • If no query is provided, ask the user what they want to search for.
   • Identify key terms, agent names, project IDs, or date ranges mentioned.

2. Search messages

   • Call search_messages with the query terms to find relevant conversations.
   • Note the channel, sender, timestamp, and content of matching messages.
   • If results are sparse, try alternative phrasings or broader terms.

3. Check project state

   • If the query relates to a specific project, call get_state with the project ID.
   • If the query mentions status, progress, or blockers, check relevant state fields.
   • Call get_state_history if the query is about how something changed over time.

4. Check task history

   • Call list_tasks and filter results relevant to the query.
   • Look for tasks matching the topic, assigned to mentioned agents, or in the queried status.

5. Check channel history (if needed)

   • If message search is insufficient, call get_history for specific channels that are likely relevant.

6. Compile results

   • Organize findings into a structured response:

   Search Results for: "<query>"
   
   Messages:
   - [timestamp] #channel @sender: relevant excerpt
   - ...
   
   Project State:
   - [project_id] field: value (if relevant)
   
   Tasks:
   - [task_id] title — status — assignee
   
   Summary:
   - Brief synthesis of what was found and how it answers the query
   

7. Highlight gaps

   • If the search did not find sufficient information, note what is missing.
   • Suggest who to ask or where to look next.

$ARGUMENTS

• /search-context <query>
• Example: /search-context authentication implementation decision
• Example: /search-context what did CEO say about launch timeline