--- name: mcs-context description: "Use this skill to pull M365 history about a customer or company before research. Searches emails, meetings, documents, Teams, and people via WorkIQ MCP to compile a customer context file. Use when starting a new project, when the user mentions a customer name, or when you need business context for architecture decisions." --- # MCS Customer Context Pull Use WorkIQ MCP to search all M365 history for a customer/company name and compile findings into structured files for use in research and architecture. ## Input Provide customer/company name and optionally project name: - `/mcs-context CDW` → searches for "CDW", uses "CDW" as project name - `/mcs-context CDW ProjectName` → searches for "CDW", saves to Build-Guides/ProjectName/ ## Prerequisites - **WorkIQ CLI installed**: `workiq --version` (must be authenticated) - **WorkIQ MCP configured**: In `.claude/settings.json` → mcpServers → workiq - **Microsoft tenant authenticated**: WorkIQ must have completed interactive auth with the target tenant - **M365 Copilot license**: Required on the authenticated account If WorkIQ is not authenticated, guide the user: ``` WorkIQ needs one-time authentication. Run this in a terminal: workiq ask -q "What is my name?" Complete the browser sign-in, then re-run /mcs-context. ``` ## Process ### Step 1: Ensure Project Folder Exists Check/create `Build-Guides/[ProjectName]/`. If it doesn't exist, create it. ### Step 2: Run Targeted WorkIQ Queries Run these queries using WorkIQ MCP tools (`mcp__workiq__*`). If MCP tools are not available, fall back to `workiq ask -q "query"` via Bash. **IMPORTANT**: Run queries in parallel where possible for speed. Each query targets a different M365 data source. #### Query 1: Broad Overview ``` "Give me a comprehensive summary of everything related to [CUSTOMER]. Include any projects, discussions, decisions, pain points, and key people involved." ``` #### Query 2: Email History ``` "Find all emails mentioning [CUSTOMER]. Summarize the key threads, decisions made, action items, and any requirements or pain points discussed. Include dates and participants." ``` #### Query 3: Meetings & Transcripts ``` "Find all meetings about [CUSTOMER] or with [CUSTOMER] participants. Summarize meeting outcomes, decisions, action items, and any technical discussions. Include dates." ``` #### Query 4: Documents & Files ``` "Find all documents, presentations, and files about [CUSTOMER] in SharePoint and OneDrive. List each file with its name, location, date, and a brief description of its content." ``` #### Query 5: Teams Messages ``` "Find all Teams messages and channel discussions mentioning [CUSTOMER]. Summarize key conversations, decisions, and any blockers or issues raised." ``` #### Query 6: People & Stakeholders ``` "Who are the key people working with [CUSTOMER]? Include internal team members and any external contacts mentioned. List their roles and involvement." ``` #### Query 7: SDR & Requirements (targeted) ``` "Find any Solution Discovery Report, requirements document, use case document, or agent specification related to [CUSTOMER]. Summarize the key requirements, use cases, and technical details." ``` #### Query 8: Recent Activity (last 30 days) ``` "What has happened with [CUSTOMER] in the last 30 days? Any recent emails, meetings, documents, or decisions?" ``` ### Step 3: Compile Results into customer-context.md Organize all findings into a structured narrative: ```markdown # Customer Context: [CUSTOMER] > Generated on [DATE] via WorkIQ M365 search > Tenant: [authenticated tenant] > Queries run: [count] ## Executive Summary [2-3 paragraph synthesis of everything found — who is this customer, what are we doing with them, what stage are they at, what are the key pain points and requirements] ## Key Stakeholders ### Internal (Our Team) | Name | Role | Involvement | |------|------|-------------| | [Name] | [Role] | [How they're involved] | ### External (Customer Side) | Name | Role | Involvement | |------|------|-------------| | [Name] | [Role] | [How they're involved] | ## Engagement History ### Timeline [Chronological summary of key events — first contact, meetings, decisions, milestones] ### Key Decisions Made - [Decision 1] — [date, context] - [Decision 2] — [date, context] ### Open Action Items - [ ] [Action item] — [owner, due date if known] ## Requirements & Use Cases Found ### From SDR/Requirements Documents [Summarize any formal requirements found] ### From Conversations (Emails/Meetings/Teams) [Summarize requirements discussed informally] ### Pain Points Mentioned - [Pain point 1] — [where it was mentioned] - [Pain point 2] — [where it was mentioned] ## Documents & Artifacts | Document | Type | Location | Date | Key Content | |----------|------|----------|------|-------------| | [Name] | [SDR/Spec/Deck/etc.] | [SharePoint path] | [Date] | [What it contains] | ## Recent Activity (Last 30 Days) [Summary of recent engagement — momentum indicator] ## Gaps & Unknowns [What WASN'T found — topics with no data, unanswered questions, areas where M365 had no results] ``` ### Step 4: Generate customer-interactions.csv Create a structured timeline CSV: ```csv "date","type","participants","summary","source" "2026-01-15","email","Dennis Kim, John Smith","Discussed CDW agent requirements — need Jira integration","Email thread: CDW Agent Planning" "2026-01-20","meeting","Dennis Kim, Sarah Lee, CDW team","SDR walkthrough — identified 3 agent use cases","Meeting: CDW Discovery Session" "2026-01-22","document","Dennis Kim","SDR v1 uploaded to SharePoint","SharePoint: /sites/Customers/CDW/SDR-v1.docx" "2026-02-01","teams","Dennis Kim, PM team","Flagged Jira on-prem as blocker for Agent 1","Teams: #customer-cdw channel" ``` ### Step 5: Write Files 1. Write `Build-Guides/[ProjectName]/customer-context.md` 2. Write `Build-Guides/[ProjectName]/customer-interactions.csv` ### Step 6: Present Summary and Guide Next Step ``` ## Customer Context Complete: [CUSTOMER] **Files created:** - Build-Guides/[ProjectName]/customer-context.md — Full narrative context - Build-Guides/[ProjectName]/customer-interactions.csv — Structured timeline ([N] interactions) **Summary:** - [Key finding 1] - [Key finding 2] - [Key finding 3] **SDR/Requirements found:** [Yes — list files / No — will need manual input] **Recommended next step:** - Run `/mcs-research [ProjectName]` → reads docs + customer-context.md, identifies agents, full enrichment ``` ## How Research Uses This When `/mcs-research` runs, it checks for `customer-context.md` in the project folder: 1. **If found**, incorporate into requirements validation: - **Problem & Priority** ← Executive Summary + Pain Points - **Users & Scenarios** ← Stakeholders + Requirements/Use Cases - **Requirements** ← Requirements Found + Documents - **Scope** ← Decisions Made + Gaps & Unknowns - **Open Questions** ← Gaps & Unknowns 2. **This pre-fills 60-80% of the research phase** — less manual validation needed ## WorkIQ Query Tips - **Be specific with company names** — "CDW" not "the customer" - **Use time ranges** when results are too broad — "in the last 6 months" - **Follow up on interesting threads** — if a query mentions an important meeting, ask WorkIQ for details about that specific meeting - **Cross-reference**: If emails mention a document, query for that document specifically - **No results ≠ no data** — WorkIQ can only access data the authenticated user has permissions to see. Note permission gaps in the Gaps section. ## Error Handling | Error | Resolution | |-------|-----------| | WorkIQ not authenticated | Guide user to run `workiq ask -q "test"` in terminal for interactive auth | | MCP server not available | Fall back to `workiq ask -q "query"` via Bash | | No results for customer | Try alternate names, abbreviations, domain names. Note "no M365 data found" | | Timeout on queries | Run queries sequentially instead of parallel. Reduce scope with time ranges | | Rate limiting | Add 5-second delays between queries | ## Output Creates: - `Build-Guides/[ProjectName]/customer-context.md` — Complete narrative context - `Build-Guides/[ProjectName]/customer-interactions.csv` — Structured timeline These files persist and can be referenced by `/mcs-research` and during builds.