Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →microsoft/ff-oce-dashboard
Generate the OCE shift status dashboard.
Passed 13 of 18 test cases with the skill, and 8 of 18 without it.
and 12 more in the full report
10 fixed7 passed either way3 still failing3 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 10 test cases the skill fixed. Same model, same prompt, with and without the skill.
When the user selects the choice to write the OCE shift dashboard to a local Markdown file, what exact file naming pattern should be used for the generated document?
| Checked | Without skill | With skill |
|---|---|---|
| Formats filename matching pattern oce-dashboard-<YYYYMMDD-HHmmss>.md. | Missed | Passed |
The instructions your model reads when the skill loads.
Gathers data from multiple MCP servers in parallel and presents a consolidated shift status dashboard. Best-effort — partial results are always shown.
Before gathering any data, use the ask_user tool:
Question: "Would you like me to generate a shift status dashboard? This will query IcM, ADO, Kusto, Teams, and WorkIQ and may trigger auth prompts. Make sure you're connected to VPN first — it's required for Kusto and other internal services."
Choices: "Yes", "Yes, and write it to a file", "No"
oce-dashboard-<YYYYMMDD-HHmmss>.md in the current working directory instead of printing it to the console. Just confirm the filename when done.Use the task tool with mode: "background" for each. Instruct each agent to always terminate — either return its results or respond with "FAILED: <reason>" if the tool call errors or auth fails. Agents must never hang or retry indefinitely.
CRITICAL: Background agents do not inherit the agent prompt context. Each sub-agent prompt must be self-contained with all IDs, parameters, and tool-call details it needs. Use the templates below — do not construct prompts from memory.
Call icm-search_incidents_by_owning_team_id for each team: 98481 (FF Hot), 149377 (Fluid Framework Client), 98313 (Azure Fluid Relay Client). Only include Active and Mitigated incidents — exclude Resolved. Report ID, Sev, Title, Status, Team, Created date.
Call ado-pipelines_get_builds for each pipeline definition (12 = Build, 56 = E2E, 63 = Stress) with project: "internal", branchName: "refs/heads/main", statusFilter: "Completed", top: 3, queryOrder: "FinishTimeDescending". Result codes: 2 = ✅, 4 = ⚠️, 8 = ❌. Report build ID, result, finish time, and overall trend.
Call ado-office-pipelines_get_builds with project: "OC", definitions: [29163], top: 5, queryOrder: "FinishTimeDescending". Important: Use the ado-office MCP server tools (NOT the default ado tools) — this pipeline is in the office ADO org, not fluidframework. Result codes: 2 = ✅, 4 = ⚠️, 8 = ❌. Report build ID, result, finish time, branch, and build number. Flag any failures — a failing integration pipeline means the next FF bump to Loop will break.
Do not load the ff-oce-kusto skill. Call kusto-kusto_query with cluster_uri: "https://kusto.aria.microsoft.com", database: "6a8929bcfc6d44e9b13fee392ada9cf0", and this query:
test cases passed, vs 8 / 18 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.