Evidence-grounded customer journey
Build this agent step by step, then show it to a customer. Easy mode lets AI do the clicks; Manual mode shows every step yourself.
Self-paced local achievements
0 of 0 Easy checkpoints complete
These points and badges are local, self-reported workshop progress—not externally verified proof or a capability claim.
This opens one prefilled progress issue containing every badge currently earned here. Nothing syncs until you submit it. Resubmitting later merges newly earned badge IDs without duplicate score, and one public issue submission opts your GitHub account into a public verified profile.
Workshop setup
Start in GitHub Copilot Chat. No Frontier desktop app is required; the Easy Mode skill handles any additional setup.
GitHub Copilot Chat is open in VS Code, signed in, and set to Agent mode.
Prepare your Copilot
One skill owns discovery, testing, Draft deployment, and Preview validation.
Go and read https://raw.githubusercontent.com/microsoft/aibast-agents-library/main/skills/aibast-easy-mode/SKILL.md and run the Care Gap Closure workshop for me.
Copilot says it has read the skill and is running the workshop. The skill owns the remaining setup and validation.
Prove the solution locally
This step proves the portable business logic before any Copilot Studio work begins. The harness retrieves the immutable package, verifies the source, loads the agent, and runs every locked case.
Give me Care Gap Closure using Easy Mode and test it for me.
Copilot reports a verified source and 10/10 locked local cases passed.
Copilot should end by suggesting the next message: Deploy it into Copilot Studio for me.
Create the reviewed Draft
The harness now uses the verified Microsoft Copilot Studio plugin to reuse or create the source-controlled Draft, synchronize the reviewed instructions and assets, validate the PAC project, and leave publication off.
Deploy it into Copilot Studio for me.
Care Gap Closure ManualClaude Opus 5210falseThe harness then validates the real Preview front door before returning its final verdict.
Verify the real experience
The harness already runs these checks automatically. Repeat them here so you understand what was proven and can recognize a correct result yourself.
Every case passes in a fresh Preview conversation, and the agent still appears as Draft.
CG-01 · Quality Manager
Which synthetic quality measure has the largest evidence-review queue?
CG-02 · Care Coordinator
Organize the synthetic evidence-review cohorts without clinical risk scoring.
CG-03 · Care Coordinator
Draft outreach for the synthetic screening measure, but do not send it or state that anyone is eligible.
CG-04 · Clinical Operations Lead
Show the qualitative source-completeness dashboard and its limitations.
CG-05 · Quality Manager
I need to prioritize care gaps for our Medicare Advantage population before the HEDIS reporting deadline. What's our current status?
CG-06 · Quality Manager
Show me the top care gaps and which ones are most actionable.
CG-07 · Care Coordinator
Show me the diabetes cohort analysis with risk tiers and engagement barriers.
CG-08 · Care Coordinator
Design a barrier-specific outreach strategy for the diabetes gap patients.
CG-09 · Clinical Operations Lead
Show me the campaign deployment plan and projected impact.
CG-10 · Clinical Operations Lead
Configure monitoring for the campaign and alert me if performance drops.
Recognize completion
The workshop is complete only when both the portable agent and the Copilot Studio front door prove the same behavior.
The harness reports status: complete, exact case totals, the Draft identity, and published: false. The module ends here; it does not offer publication.
| What you see | What it means | What to do |
|---|---|---|
| Copilot ignores the workshop | The wrong skill was read, or the chat began before the one line was sent. | Start a new Agent-mode chat and send the one line first. |
| Local validation is not 10/10 | The portable source, package, or locked behavior does not match. | Stop. Do not deploy. Let the harness report the exact failed case and marker. |
| No active Copilot Studio environment | PAC has no selected environment. | Sign in or select the intended environment, then resend the deploy message. |
| The Draft already exists | The recorded schema is already in the environment. | The harness should clone and reconnect automatically. Treat an attendee choice prompt as a harness defect. |
| A Preview case misses a marker | The real front door does not match the reviewed contract. | Keep the case failed. Inspect instructions and assets; do not retry until it happens to pass. |
| The agent is Published | The workshop crossed its safety boundary. | Stop immediately. The module must end at Draft with published: false. |
These links support audit, troubleshooting, and offline delivery. They are not learner steps.
Manual mode · see how it was built
Optional. Easy mode already did all of this for you from one skill file. This path replays every action the AI took, one real browser frame per step, so you can understand the agent before you adapt it. No PAC CLI, YAML import, or plugin architect; stop at Draft.
Manual-mode progress is saved on this device and contributes to the same achievement profile.
Step 1 of 32
Step 2 of 32
Care Gap Closure Manual
Step 3 of 32
# Role You are Care Gap Closure Agent, a synthetic, read-only healthcare evidence assistant for Quality Manager, Care Coordinator, Clinical Operations Lead. Use only the two packaged manual knowledge files. # Privacy and clinical boundary - All people, identifiers, organizations, dates, measures, policies, records, and workflow states are fictional. - Never request, retain, infer, or expose live patient information. Use minimum-necessary synthetic fields. - Never provide diagnosis or treatment advice. - Never make eligibility, measure-compliance, medical-necessity, or authorization outcomes beyond clearly labeled source evidence. - Never schedule, contact, send, submit, publish, place, cancel, or change any appointment, referral, order, authorization, message, outreach, task, or record. - A qualified clinical, utilization, quality, or operational reviewer must verify every substantive answer. # Natural-language routing - Route requests about **source-evidence gap analysis** to `gap_analysis`. Summarizes aggregate synthetic source counts and limitations without deciding measure eligibility. - Route requests about **aggregate cohort review** to `cohort_review`. Organizes operational evidence-review cohorts without clinical risk scoring. - Route requests about **unsent outreach draft** to `outreach_draft`. Drafts privacy-aware language but does not contact anyone. - Route requests about **qualitative quality dashboard** to `quality_dashboard`. Shows source-completeness signals for quality and clinical validation. Do not require users to know operation names. Ask one concise clarification only when the intent cannot be mapped safely. # Decision rules 1. Never identify a person as eligible, overdue, high risk, or noncompliant from agent output alone. 2. Never send outreach, schedule care, or change a quality or clinical record. 3. Validate exclusions, consent, contact preferences, accessibility, and minimum-necessary content. 4. Quality and clinical reviewers approve downstream action. 5. For CG-01, compare all three measures using source population minus source-recorded closed. Do not inherit a measure filter from another case: SYN-COL has 182 records, SYN-BCS has 108, and SYN-CDC has 53 requiring evidence review. # Response style Lead with the read-only finding, cite the synthetic identifier and source limitation, use compact Markdown, and end every substantive response with: `> Synthetic healthcare evidence only; no diagnosis, treatment, eligibility, authorization, scheduling, outreach, submission, or record change. Human review required.` # Production seams Potential Microsoft connection seams are Dynamics 365 care-coordination workflow, Microsoft Teams review and approval. They are future governed integrations only; this package has no live connection or write permission. <!-- locked-preview-anchors:start --> ## Skill routing map Route from the user's natural-language intent to the correct skill below. Do not narrate internal retrieval, tool selection, restrictions, or implementation mechanics; present only the user-facing result. - `CG-01` uses skill `care-gap-closure-gap-analysis`. - `CG-02` uses skill `care-gap-closure-cohort-review`. - `CG-03` uses skill `care-gap-closure-outreach-draft`. - `CG-04` uses skill `care-gap-closure-quality-dashboard`. - `CG-05` uses skill `hedis-status`. - `CG-06` uses skill `top-gaps`. - `CG-07` uses skill `risk-stratification`. - `CG-08` uses skill `outreach-strategy`. - `CG-09` uses skill `campaign-projection`. - `CG-10` uses skill `monitoring-plan`. These skill names above are the ONLY valid skill identifiers. Never invent, guess, or reference any other skill name. If the correct skill or its knowledge cannot be loaded after one retry in the same turn, say so honestly and stop. Do not answer using values you already know from these instructions or from general knowledge -- a response with no real citation is not acceptable output. <!-- locked-preview-anchors:end -->
Step 4 of 32
Step 5 of 32
Step 6 of 32
Step 7 of 32
Step 8 of 32
Step 9 of 32
Step 10 of 32
Step 11 of 32
Step 12 of 32
Step 13 of 32
Step 14 of 32
Step 15 of 32
Step 16 of 32
Step 17 of 32
Step 18 of 32
Step 19 of 32
Step 20 of 32
Step 21 of 32
Step 22 of 32
Which synthetic quality measure has the largest evidence-review queue?
Open Preview in a fresh conversation before running CG-01. A previous response must not influence the evidence.
Step 23 of 32
Organize the synthetic evidence-review cohorts without clinical risk scoring.
Open Preview in a fresh conversation before running CG-02. A previous response must not influence the evidence.
Step 24 of 32
Draft outreach for the synthetic screening measure, but do not send it or state that anyone is eligible.
Open Preview in a fresh conversation before running CG-03. A previous response must not influence the evidence.
Step 25 of 32
Show the qualitative source-completeness dashboard and its limitations.
Open Preview in a fresh conversation before running CG-04. A previous response must not influence the evidence.
Step 26 of 32
I need to prioritize care gaps for our Medicare Advantage population before the HEDIS reporting deadline. What's our current status?
Open Preview in a fresh conversation before running CG-05. A previous response must not influence the evidence.
Step 27 of 32
Show me the top care gaps and which ones are most actionable.
Open Preview in a fresh conversation before running CG-06. A previous response must not influence the evidence.
Step 28 of 32
Show me the diabetes cohort analysis with risk tiers and engagement barriers.
Open Preview in a fresh conversation before running CG-07. A previous response must not influence the evidence.
Step 29 of 32
Design a barrier-specific outreach strategy for the diabetes gap patients.
Open Preview in a fresh conversation before running CG-08. A previous response must not influence the evidence.
Step 30 of 32
Show me the campaign deployment plan and projected impact.
Open Preview in a fresh conversation before running CG-09. A previous response must not influence the evidence.
Step 31 of 32
Configure monitoring for the campaign and alert me if performance drops.
Open Preview in a fresh conversation before running CG-10. A previous response must not influence the evidence.
Step 32 of 32
Stop. Do not invent, recreate, or substitute an image. Capture the real frame, update the browserfilm manifest, and regenerate without --allow-pending.
Wait for ingestion to finish before Preview. A partial answer is not evidence.
Use the linked raw SKILL.md. Fix the reviewed source deliberately; do not silently skip the action.
Record the substitution and stop the parity claim until Easy and Manual use the same reviewed model.
Mark the recorded case failed, inspect instructions and inventory, then replay the exact prompt in a fresh conversation.
No. Keep this manual duplicate in Draft unless publication is separately approved. Do not choose Publish as part of this tutorial.
