aurelienclere-365/ado-cowork-plugin/ado-pm-reporting

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Use when the user asks for project management reports, release readiness, sprint health, team velocity, backlog grooming, stakeholder updates, or retrospective publishing. "sprint health report", "go/no-go for the release", "release readiness check", "velocity trend for the last 4 sprints", "how much can we commit next sprint", "weekly status update for stakeholders", "backlog grooming — unrefined stories", "items without story points or acceptance criteria", "what slipped from the sprint", "cross-project visibility on a feature", "publish retrospective notes to the wiki", "committed vs completed points", "are we on track for the release", "open P1 bugs before go-live", "team capacity vs commitment", "PM dashboard", "prepare the project committee deck", "create a PowerPoint for the steering committee", "generate a presentation from ADO data", "build a slide deck for the sprint review", "make a PowerPoint for the release committee", "project committee presentation"

Install

npx vskill@latest install aurelienclere-365/ado-cowork-plugin/ado-pm-reporting

SKILL.md

ADO PM Reporting

Composite project management workflows that combine work items, sprint data, pipelines, test results, and wiki publishing into executive-ready reports and PM artefacts.

This skill orchestrates multiple underlying MCP tool groups so the user does not need to know which individual tool to call — describe the PM need and the skill assembles the right calls in sequence.

When to use

  • User wants a sprint health report (committed vs completed, slippage, blockers)
  • User wants a go/no-go / release readiness check (open P1 bugs, build status, test pass rate)
  • User wants a velocity trend across past sprints to plan the next commitment
  • User wants a weekly stakeholder update (progress, pipeline health, blockers)
  • User wants to groom the backlog (find unrefined, unestimated, or unassigned items)
  • User wants to set up a new sprint with capacity and carry-over items
  • User wants to publish a retrospective to the wiki
  • User needs cross-project visibility on a topic (e.g. duplicate initiatives)

MCP tools used

Tool groupToolsPurpose
Work Itemswit_query, wit_backlog, wit_work_item_writeQuery and update items for reports
Sprintswork, work_iteration_write, work_capacity_writeIteration data and capacity
Pipelinespipelines_buildBuild health for release checks
Test Planstestplan_show_test_results_from_build_idTest pass rate for release checks
Searchsearch_workitemCross-project item discovery
Wikiwiki_upsert_pagePublish retrospectives and PM notes

Workflow A — Sprint Health Report

Triggers: "sprint health report", "sprint status", "what slipped", "committed vs completed"

  1. Call work(project, team) — get the active or named iteration name and date range.
  2. Call wit_query(wiql) for all items in that iteration; group by state

(Done/Closed = completed, Active/In Progress = in flight, New = not started).

  1. Sum story points per group. Calculate completion rate = completed ÷ committed × 100.
  2. Identify slipped items: items still in Active or New state at sprint end (or near end).
  3. Present Sprint Health output. Offer to move slipped items to next sprint.
Output — Sprint Health

{Team} — Sprint {N} Health Report

MetricValue
Committed points42
Completed points35
Completion rate83%
Items completed11 / 14
Slipped items3

Slipped items:

IDTypeTitlePointsAssignee
1042StoryAdd MFA to login flow5Alice Martin
1058BugSession token not refreshed2(unassigned)

Workflow B — Go/No-Go Release Readiness

Triggers: "go/no-go", "release readiness", "are we ready to release", "open P1 bugs before go-live"

  1. Call wit_query for open bugs with Priority = 1 in the project. Flag count and owners.
  2. Call pipelines_build(project, null, releaseBranch) — get latest build status on the

release branch.

  1. Get that build's ID; call testplan_show_test_results_from_build_id(buildId) —

summarise pass/fail/skipped counts and failing test names.

  1. Assemble a Go/No-Go table. Apply rule: NO-GO if any P1 bug is open OR build is

failing; WARN if test pass rate < 98%.

Output — Go/No-Go

{Version} Release — Go/No-Go Assessment

CheckStatusDetail
Open P1 bugs⚠ 2 open#1058 unassigned, #1063 Alice
Release branch build✅ PassingBuild #512 — 2h ago
Test pass rate✅ 98.9%3 failures (integration)
VerdictNO-GOResolve P1 bugs before release

Workflow C — Velocity Trend & Sprint Commitment

Triggers: "velocity trend", "how much can we commit", "team velocity", "capacity vs velocity"

  1. Call work(project, team) to list all past iterations (last N sprints as requested,

default 4).

  1. For each past iteration, call wit_query for items with state Done/Closed in that

iteration path; sum story points → velocity for that sprint.

  1. Calculate average velocity. Recommend commitment range = average ± 10%.
  2. If capacity data is available (work_capacity_write was previously set), compare

available hours vs historical throughput per point.

Output — Velocity Trend

{Team} — Velocity (last 4 sprints)

SprintCommittedCompletedVelocity
Sprint 9383636
Sprint 10403838
Sprint 11454040
Sprint 12423535
Average——37
Recommended Sprint 13 commitment: 34–39 points.

Workflow D — Stakeholder Weekly Update

Triggers: "weekly update", "stakeholder report", "weekly status", "progress summary"

  1. Call wit_query for items moved to Done/Closed in the last 7 days across the project.
  2. Call pipelines_build for the project's key pipelines — summarise pass/fail.
  3. Call wit_query for items tagged "blocker" or in Blocked state (or flagged).
  4. Draft a structured update using the three data sets.
Output — Stakeholder Update
{Project} — Weekly Update (Week of {date})
Completed ({N} items): brief list of closed work items
Pipeline health: per-pipeline pass/fail summary
Blockers: open blocked items with owner

Workflow E — Backlog Grooming

Triggers: "backlog grooming", "unrefined stories", "no story points", "no acceptance criteria", "unassigned backlog", "items not in a sprint"

  1. Build WIQL query filtering: Type = User Story, State = New,

IterationPath = project root (not assigned to a sprint), StoryPoints = null.

  1. Call wit_query(wiql) — return results ordered by Priority.
  2. Present the Grooming Queue table.
  3. Offer bulk actions: "Set story points on #X to N", "Assign #X to Sprint 13".
Output — Grooming Queue
PriorityIDTitleAssigned ToNotes
11081Implement SSO for mobile app(unassigned)No points, no sprint
21085Add export to Excel on reports(unassigned)No points, no sprint

Workflow F — New Sprint Setup

Triggers: "set up sprint", "create next sprint", "plan Sprint N", "new iteration with capacity"

  1. Call work_iteration_write(project, name, startDate, endDate) — create the iteration.
  2. Call work_capacity_write for each team member with hours/day and days off.
  3. If the user mentions carry-over items, call wit_work_item_write to update iteration

path on each item ID.

  1. Confirm iteration, total available hours, and moved items.

Workflow G — Retrospective Publishing

Triggers: "publish retrospective", "retrospective notes", "retro to wiki", "sprint retro"

  1. Extract four sections from the user's input:
  • What went well
  • What to improve
  • Team agreements (new rules/norms)
  • Action items (owner + due date)
  1. Format as a structured Markdown page using the standard retro template (see below).
  2. Call wiki_upsert_page(project, wikiId, "Retrospectives/Sprint-N", content, comment).
  3. Return the published page URL.
Retrospective template
markdown
# Sprint {N} Retrospective — {Team}
**Date:** {date} **Facilitator:** {name}
## What went well
-...
## What to improve
-...
## Team agreements
-...
## Action items
| Action | Owner | Due |
|---|---|---|
| ... | ... | ... |

Workflow I — Project Committee PowerPoint Deck

Triggers: "prepare the committee deck", "create a PowerPoint", "slide deck for the steering committee", "project committee presentation", "sprint review deck", "generate slides from ADO data"

This workflow pulls live ADO data, assembles it into a structured slide outline, and hands it to Microsoft 365 Copilot's native PowerPoint generation (create_presentation) to produce a .pptx file — no manual copy-paste between ADO and PowerPoint needed.

Availability: PowerPoint generation requires Microsoft 365 Copilot (Option C —
M365 connector deployment). In VS Code Copilot (Option B), the skill produces the full
slide outline as structured Markdown that you can paste into PowerPoint or Copilot for
PowerPoint manually.
Step 1 — Gather ADO data

Run the relevant workflows first (A–H above) based on what the committee needs:

Meeting typeWorkflows to run first
Sprint reviewA (sprint health) + C (velocity)
Release / go-live committeeB (go/no-go) + A (sprint health)
Quarterly steering committeeC (velocity trend) + D (stakeholder update) + E (backlog)
Retrospective sessionG (retrospective data)
Step 2 — Assemble the slide outline

Structure the collected data into the slide deck blueprint below. Fill each section from the workflow outputs:

Slide 1 — TITLE
{Project name} — {Meeting type}
{Date} | Presented by {PM name}
Slide 2 — AGENDA
1. Sprint / Release summary
2. Key achievements
3. Risks & blockers
4. Next sprint / next steps
5. Decisions needed
Slide 3 — SPRINT / RELEASE SUMMARY (from Workflow A or B)
• Committed: {N} pts Completed: {N} pts ({%} rate)
• Build status: {pass/fail} Test pass rate: {%}
• Slipped items: {N}
Slide 4 — KEY ACHIEVEMENTS (top completed items from Workflow D)
• {Title of completed work item 1}
• {Title of completed work item 2}
• {Title of completed work item 3}
(limit to top 5 — choose highest priority / most visible)
Slide 5 — RISKS & BLOCKERS
• Open P1 bugs: {list from Workflow B}
• Blocked items: {list from Workflow D}
• Each row: ID | Title | Owner | Mitigation
Slide 6 — VELOCITY & FORECAST (from Workflow C)
• Velocity chart: last 4 sprints (table form if chart not available)
• Next sprint recommendation: {N}–{N} points
• Capacity: {total available hours}
Slide 7 — NEXT SPRINT / NEXT STEPS (from Workflow F or backlog)
• Sprint {N}: {start}–{end}
• Committed items (top 5 by priority)
• Capacity allocated: {N}h
Slide 8 — DECISIONS NEEDED
• {Item requiring committee decision 1}
• {Item requiring committee decision 2}
(extract from blocked items, slipped high-priority stories, or user input)
Slide 9 — THANK YOU / Q&A
Step 3 — Generate the PowerPoint

In M365 Copilot (Option C): Pass the assembled outline to Copilot's create_presentation capability:

"Create a PowerPoint presentation from this outline using our corporate template."

Copilot will produce a .pptx file saved to OneDrive / SharePoint, which you can share directly in the meeting invite.

In VS Code Copilot (Option B): The skill outputs the outline as formatted Markdown. Copy it and:

  • Open PowerPoint → Copilot sidebar → "Create presentation from outline" → paste
  • Or use Copilot for Microsoft 365 in PowerPoint directly
Output — Slide outline (Markdown preview)
# Platform Project — Release Committee | June 2026
## Sprint 12 Summary
- Committed: 42 pts | Completed: 35 pts | Rate: 83%
- Build: ✅ Passing | Tests: 98.9% pass
- Slipped: 3 items (moving to Sprint 13)
## Key Achievements
- Redis cache layer deployed to staging
- API rate limiting implemented
- Onboarding flow redesign — UX sign-off complete
## Risks & Blockers
| # | Title | Owner | Mitigation |
|---|---|---|---|
| 1058 | Session token not refreshed | Alice Martin | Fix in progress — due June 10 |
| 1063 | Null ref in OrderService | (unassigned) | Needs owner assignment |
## Velocity & Forecast
| Sprint | Velocity |
|---|---|
| Sprint 9 | 36 | Sprint 10 | 38 | Sprint 11 | 40 | Sprint 12 | 35 |
Average: 37 pts — Recommended commitment Sprint 13: 34–39 pts
## Next Sprint (Sprint 13 — June 9–22)
- Top 3 committed items: [SSO mobile, Export Excel, Notification prefs]
- Capacity: 124h available
## Decisions Needed
- Assign owner for bug #1063 — P1, currently unassigned
- Approve scope reduction: defer dark mode (#1092) to Sprint 14?

Workflow H — Cross-project Feature Visibility

This is the start of the file. Read the full SKILL.md on GitHub.

Security

  • Security scanPassed, 95 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/aurelienclere-365/ado-cowork-plugin/ado-pm-reporting/badge)](https://verified-skill.com/skills/aurelienclere-365/ado-cowork-plugin/ado-pm-reporting)

Something wrong with this skill or its page? Tell us.