From c2dc40e1003371b83a0bb5412c71da14c8840962 Mon Sep 17 00:00:00 2001 From: gurret lionel Date: Thu, 30 Jul 2026 11:41:25 +0200 Subject: [PATCH] Remove the dry-run step from App Modernization Challenge 1 Challenge 1 asked participants to "prompt it to modernize a sample" and confirm the agent stops at Gate 1 and Gate 2, but no application is available at that point: the PhotoAlbum samples are only forked in Challenge 2. The step could not be completed as written. Remove the dry-run action and its matching success criterion from both the challenge and the walkthrough. Challenge 1 now stops at authoring and configuring the agent, skill, and MCP files. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --- .../03_GHCPAppModernization/challenges/challenge-01.md | 2 -- .../walkthrough/challenge-01/solution-01.md | 8 -------- 2 files changed, 10 deletions(-) diff --git a/03-Azure/01-01-App Innovation/03_GHCPAppModernization/challenges/challenge-01.md b/03-Azure/01-01-App Innovation/03_GHCPAppModernization/challenges/challenge-01.md index d6b38f28..01227af0 100644 --- a/03-Azure/01-01-App Innovation/03_GHCPAppModernization/challenges/challenge-01.md +++ b/03-Azure/01-01-App Innovation/03_GHCPAppModernization/challenges/challenge-01.md @@ -22,7 +22,6 @@ Before pointing AI at a real .NET or Java application (challenges 2 and 3), you * **Author a Custom Agent** — copy [templates/agents/modernization.agent.md](../templates/agents/modernization.agent.md) to `.github/agents/.agent.md` in a target repo, replace every `{{PLACEHOLDER}}`, and tighten the tool allow-list so the assessment phase stays read-only (least privilege). * **Author a Skill** — copy [templates/skills/modernization-skill/SKILL.md](../templates/skills/modernization-skill/SKILL.md) to `.github/skills//SKILL.md`, and write an explicit `WHEN:` trigger description plus at least one transformation-rules table. Optionally use the [skill-creator](../templates/skills/skill-creator/SKILL.md) meta-skill to iterate. * **Configure MCP** — copy [templates/mcp/mcp.json](../templates/mcp/mcp.json) to `.vscode/mcp.json`, then run **MCP: List Servers** and confirm the `appmod-*` tools (registered by the App Modernization extensions) appear in the agent's tool picker. -* **Dry-run the loop** — select your Custom Agent in Copilot Chat and prompt it to modernize a sample. Confirm it **stops at Gate 1 (assessment)** and **Gate 2 (plan)** before editing any code. * Compare your work against the pre-tailored references shipped in [templates/](../templates/) (`dotnet-modernization.agent.md`, `dotnet-upgrade`, `java-modernization.agent.md`, `java-upgrade`). ## Success criteria @@ -31,7 +30,6 @@ Before pointing AI at a real .NET or Java application (challenges 2 and 3), you * A valid Custom Agent (`.agent.md` with YAML frontmatter) exists, with a phased, gated workflow and a least-privilege tool list. * A valid Skill (`SKILL.md`) exists with an explicit `WHEN:` trigger description and at least one rules table. * `.vscode/mcp.json` is configured and **MCP: List Servers** resolves the servers; the `appmod-*` tools are visible to the agent. -* When run, the agent respects the gates — it does **not** modify code before the assessment and plan are approved. > Need the detailed, step-by-step walkthrough? See the [Challenge 1 Solution](../walkthrough/challenge-01/solution-01.md). diff --git a/03-Azure/01-01-App Innovation/03_GHCPAppModernization/walkthrough/challenge-01/solution-01.md b/03-Azure/01-01-App Innovation/03_GHCPAppModernization/walkthrough/challenge-01/solution-01.md index 727c7566..0c7eaea2 100644 --- a/03-Azure/01-01-App Innovation/03_GHCPAppModernization/walkthrough/challenge-01/solution-01.md +++ b/03-Azure/01-01-App Innovation/03_GHCPAppModernization/walkthrough/challenge-01/solution-01.md @@ -44,20 +44,12 @@ Author a reusable, gated modernization Custom Agent, package a Skill, and config 3. Open the Command Palette → **MCP: List Servers** and start/inspect the servers. 4. Confirm the `appmod-*` tool families appear in the agent's tool picker — they are registered automatically by the GitHub Copilot app modernization extensions (install the extension for your stack; no `mcp.json` entry needed). -### Dry-run the gated loop - -1. Select your Custom Agent in Copilot Chat. -2. Prompt: *"Modernize this application following your workflow."* -3. Verify the agent runs the assessment and **stops at Gate 1** with an `ASSESSMENT.md`, then **stops at Gate 2** with a `PLAN.md` — without editing any source code. -4. Approve to continue, or reject the plan once on purpose and watch the agent revise it. - ## Success Criteria - ✅ You can explain Agent vs Skill vs MCP in one sentence each - ✅ A valid `.agent.md` exists with a gated, phased workflow and a least-privilege tool list - ✅ A valid `SKILL.md` exists with an explicit `WHEN:` trigger description and a rules table - ✅ `.vscode/mcp.json` resolves via **MCP: List Servers**; `appmod-*` tools are visible to the agent -- ✅ The agent respects Gate 1 and Gate 2 and does not modify code before approval ## Learning Resources