Operations is the connective tissue of any organization — the processes, playbooks, capacity models, and performance metrics that keep everything running. This skill connects your AI assistant to UPLO's structured extraction of operational knowledge: SOPs, runbooks, capacity plans, incident postmortems, vendor SLAs, and the KPI data that tells you whether things are actually working.
Load your ops context to understand your role, team scope, and current operational priorities:
use_mcp_tool: get_identity_context
Then pull the latest on anything that might need immediate attention:
use_mcp_tool: search_knowledge query="active incidents open action items SLA breaches capacity warnings"
use_mcp_tool: get_directives
Directives for operations teams typically cover efficiency targets, cost reduction mandates, and service level commitments — knowing these frames every decision you make.
Something went wrong and you need to contain it, then learn from it.
use_mcp_tool: search_knowledge query="runbook incident response procedure for payment processing failures"
use_mcp_tool: search_knowledge query="previous incidents payment processing root cause analysis postmortem"
use_mcp_tool: search_with_context query="payment processing system dependencies upstream downstream SLA obligations"
The first search gets you the immediate playbook. The second surfaces prior incidents so you can check whether this is a recurring pattern. The context search maps system dependencies so you understand blast radius.
You need to model whether current resources can handle projected Q3 volume.
use_mcp_tool: search_knowledge query="capacity utilization rates by team department Q1 Q2 actual vs planned"
use_mcp_tool: search_knowledge query="demand forecast projections Q3 volume transaction throughput"
use_mcp_tool: search_knowledge query="hiring plan headcount approved positions open requisitions operations"
use_mcp_tool: export_org_context
The org context export gives you the current organizational structure overlaid with capacity data, making it clear where you have headroom and where you're already running hot.
search_knowledge — Your primary tool for finding SOPs, runbooks, KPI data, and process documentation. Operations data is often spread across wikis, shared drives, and ticketing systems — UPLO consolidates it into searchable structured records. Example: "order fulfillment process cycle time SLA target vs actual last 6 months"
search_with_context — Operations is all about dependencies. A process change in one area cascades through others. This tool follows those connections. Example: "upstream dependencies for the monthly close process including data feeds handoffs and approval gates"
export_org_context — Generates a snapshot of your operational structure: teams, systems, processes, and their interconnections. Use it to brief new team members or to give leadership a helicopter view of operational health.
flag_outdated — Stale runbooks are dangerous. If you encounter a procedure that references a decommissioned system, an old vendor, or a changed approval chain, flag it immediately. Example: flag a disaster recovery plan that still references the on-prem data center you migrated off of 18 months ago.
propose_update — After a process improvement, push the updated procedure back into the knowledge base. Don't let the documentation drift from reality. Example: update the customer onboarding SOP to reflect the new automated verification step.
"MTTR mean time to repair" or "NPS net promoter score customer operations" — this catches documents regardless of which form they used.flag_outdated on the stale version AND report_knowledge_gap to note the conflict so the process owner can reconcile them.log_conversation to document the rationale and expected outcomes. This creates an audit trail that's invaluable when someone later asks "why did we change this?"共 1 个版本