[workshop-update] Fix destroy-lifecycle content drift and close workshop glossary/skill gaps

The issue names five workshop Markdown files, including the shared glossary and Track 3 Lab 6. Read the cited skill and agent source files, then review the workshop passages listed under each change to confirm they match the described behavior. Done means the destroy guidance is consistent, the listed skills and agent are named in the glossary or lab text, and the stated timing and fallback paths remain intact.

[!TIP]
Your pull request is ready to create! 🎉 ✅

Everything is OK—the changes have been pushed to branch workshop-content-updates-20261019-4c1ff6efcab911b5. Please review the changes, including any protected files, before creating the pull request.

Create the pull request

The original pull request description is below.

Weekly workshop coverage pass. Re-confirms and re-applies the same content-drift fixes documented in prior open items (#414/#392 — still unmerged on main), verified fresh against current main, plus coverage for one newly-landed feature (/git-ape-skills) that neither prior pass covered.

Changes, per feature

1. Track 3 Lab 6 ("Destroy Lifecycle") — internal contradiction, fixed
workshops/track-3-platform-engineering/lab-06-destroy-lifecycle.md

Steps 1-5 described the old per-resource-group destroy flow (read RG name → inventory → delete subscription-scoped resources → delete RG), while Steps 6-8 (appended in a later pass) correctly describe the current Azure Deployment Stack flow (az stack sub delete --action-on-unmanage deleteAll). These directly contradicted each other within the same lab. Merged into one consistent stack-based narrative in Step 4 and removed the duplicated Steps 7-8.

• Persona/track: DevOps/SRE / Track 3 — the audience that runs the CI destroy workflow and needs an accurate mental model of what it does.

2. az group delete anti-pattern in cleanup sections — fixed
workshops/track-2-deploy-like-a-pro/lab-05-drift-detection.md, workshops/track-3-platform-engineering/lab-06-destroy-lifecycle.md

Both labs told attendees to clean up with a raw az group delete. .github/skills/azure-stack-destroy/SKILL.md explicitly flags this as wrong for Git-Ape-managed deployments — it skips the soft-delete purge sweep (Key Vault, Cognitive Services linger 7-90 days) and misses subscription-scope resources (role/policy assignments) a stack can own. Replaced both call sites with /azure-stack-destroy, the same primitive the CI destroy workflow uses, with an explanatory callout. Left az group delete guidance in place only for resource groups created by hand outside Git-Ape.

• Persona/track: Engineer / Track 2 and DevOps / Track 3.

3. azure-stack-deploy / azure-stack-destroy skills — glossary gap closed
workshops/shared/glossary.md

Both skills had 0 refs in workshops/ per the coverage inventory. The behaviour is already lab-covered through @git-ape's deploy/destroy flow and the fixes above, so rather than invent a new lab I added glossary entries mapping the CLI primitive attendees already see to the skill name driving it.

• Persona/track: Engineer / Track 2, DevOps / Track 3.

4. azure-requirements-gatherer agent — glossary gap closed
workshops/shared/glossary.md

0 refs by exact agent-slug match in the inventory, but extensively covered under its display name "Requirements Gatherer" in Track 1/2 labs, demo scripts, and the Track 2 deck — a grep-proxy false positive confirmed by reading the actual content. Added it to the glossary's agent table (the one agent missing there).

• Persona/track: Beginner / Track 1, Engineer / Track 2.

5. azure-resource-availability skill — named explicitly
workshops/track-1-zero-to-deploy/lab-02-first-deploy.md

The behaviour (region/runtime/provider/CAF-name checks) was already described in Lab 2 Step 3, but the skill was never named. Naming-only fix, no content invented.

• Persona/track: Beginner / Track 1.

6. azure-rest-api-reference skill — citation added
workshops/track-2-deploy-like-a-pro/lab-02-web-app-sql.md

0 refs. This is an internal correctness tool the Template Generator uses before writing ARM resources, not something attendees invoke directly in any lab today. Added a "Going further" citation rather than inventing a new lab section, since there's no attendee-facing behaviour to ground a dedicated lab in yet. Flagged as optional future work for a human to decide on (see coverage-status issue).

• Persona/track: Engineer / Track 2 (tentative).

7. git-ape-skills (community skill discovery/install) — new coverage, glossary + lab pointer
workshops/shared/glossary.md, workshops/track-3-platform-engineering/lab-06-destroy-lifecycle.md

This mechanism (.github/skills/git-ape-skills/SKILL.md, landed on main 2026-10-09) had 0 refs anywhere in workshops/ — confirmed by reading the source, not inferred from the grep proxy alone. Track 3 Lab 6's existing "Explore more skills" next-step bullet was the natural, already-present attendee touchpoint for this exact capability (search/install additional skills at workshop wrap-up), so I grounded the feature there plus a glossary entry rather than inventing a new lab. Every described behaviour (search syntax, pinned-commit install, explicit approval requirement) is verified against the skill's own SKILL.md.

• Persona/track: DevOps/SRE / Track 3 — the persona most likely to extend their own toolkit after completing the core program.

Guardrails confirmed

• No lab timing changed (Track 1 Lab 2 still 15 min, Track 2 Lab 2/5 still 15/5 min, Track 3 Lab 6 still 5 min; all track totals unchanged).
• No-Azure fallback paths untouched (Track 1 Lab 2's review-only path is intact, confirmed present in the diff).
• No rendered binaries (*.pdf, *.pptx, *.html) touched — only .md lab/glossary files.
• No deck source (*_deck.md) edited, so the L1/L11/L15 deck lint rules don't apply to this PR.
• No files outside workshops/ touched.
• Every described command/behaviour (az stack sub delete --action-on-unmanage deleteAll --bypass-stack-out-of-sync-error true, the availability-check APIs, the git-ape-skills search/install flow) is verified against the actual SKILL.md/agent.md sources, not invented.

Links

Coverage-status issue opened alongside this PR has the full gap table and persona reasoning. Does not close #414/#392 — those PRs weren't merged, so this is an independent re-application plus one net-new fix; a human should decide whether to close the older duplicates once this merges.

[!NOTE]
GitHub Actions is not permitted to create or approve pull requests in this repository.

The changes have been pushed to branch workshop-content-updates-20261019-4c1ff6efcab911b5 and are ready to review.

To fix the permissions issue, go to Settings → Actions → General and enable Allow GitHub Actions to create and approve pull requests. See also: gh-aw FAQ

From 40371b1280f65170af736e86d37b24beec83b5fa Mon Sep 17 00:00:00 2001
X-GH-AW-Base-Commit: 35806c382baa8b26ffb44705fbc1bf17659fc615
From: "github-actions[bot]" <github-actions[bot]@users.noreply.github.com>
Date: Mon, 19 Oct 2026 20:59:53 +0000
Subject: [PATCH] workshop: fix destroy-lifecycle content drift, close
 skill/agent glossary gaps

- Track 3 Lab 6 Steps 1-5 described the old per-resource-group destroy
  flow while Steps 6-8 (appended later) described the Deployment Stack
  based flow. Both directly contradicted each other. Merged into one
  consistent stack-based narrative and removed the duplicated steps.
- Track 2 Lab 5 and Track 3 Lab 6 told attendees to clean up with raw
  'az group delete', which the azure-stack-destroy SKILL.md explicitly
  flags as wrong for Git-Ape-managed deployments (misses soft-delete
  purge and subscription-scope resources). Replaced both call sites
  with /azure-stack-destroy.
- Added azure-stack-deploy, azure-stack-destroy, azure-resource-availability,
  azure-rest-api-reference, git-ape-skills, and the Requirements Gatherer
  agent to workshops/shared/glossary.md (0 refs previously per the
  coverage inventory).
- Named azure-resource-availability explicitly in Track 1 Lab 2 Step 3
  (behaviour already described, skill never named).
- Cited azure-rest-api-reference in Track 2 Lab 2 Going Further.
- Pointed Track 3 Lab 6's 'Explore more skills' next-step at the newly
  landed /git-ape-skills discovery/install mechanism, which had 0
  workshop refs.

Co-authored-by: Copilot <[email protected]>
---
 workshops/shared/glossary.md                  |  6 +++
 .../lab-02-first-deploy.md                    |  2 +-
 .../lab-02-web-app-sql.md                     |  1 +
 .../lab-05-drift-detection.md                 |  6 +--
 .../lab-06-destroy-lifecycle.md               | 54 +++++--------------
 5 files changed, 25 insertions(+), 44 deletions(-)

diff --git a/workshops/shared/glossary.md b/workshops/shared/glos
... (truncated)

Generated by Workshop Content Auto-Updater · copilot · auto · 144.4 AIC · ⌖ 18.7 AIC · ⊞ 12K · ◷