Update Learning Hub with v1.0.66-1.0.68 changes

- Add kimi-k2.7-code model (v1.0.68) to model table in building-custom-agents
- Note Claude Opus 4.8 Fast replacing Opus 4.6 Fast (v1.0.66)
- Document /pr auto self-paced loop and /pr automerge behavior (v1.0.66)
- Add dynamic skill retrieval (--dynamic-retrieval flag) to creating-effective-skills

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

website/src/content/docs/learning-hub/building-custom-agents.md

@@ -3,7 +3,7 @@ title: 'Building Custom Agents'
 description: 'Learn how to create specialized GitHub Copilot agents with custom personas, tool integrations, and domain expertise.'
 authors:
   - GitHub Copilot Learning Hub Team
-lastUpdated: 2026-07-01
+lastUpdated: 2026-07-05
 estimatedReadingTime: '10 minutes'
 tags:
   - agents
@@ -257,9 +257,12 @@ The agent can then query your database, analyze query plans, and suggest optimiz
 | Most demanding reasoning, security review | Claude Sonnet 5 *(v1.0.67+)* |
 | Complex reasoning, analysis | Claude Sonnet 4 |
 | Code generation, refactoring | GPT-4.1 |
+| Specialized code generation | kimi-k2.7-code *(v1.0.68+)* |
 | Quick analysis, simple tasks | Claude Haiku or GPT-4.1-mini |
 | Large codebase understanding | Models with larger context windows |
 
+> **Model changes in v1.0.66**: Claude Opus 4.8 Fast replaces Claude Opus 4.6 Fast. If you have agent definitions pinned to Claude Opus 4.6 Fast, update them to Claude Opus 4.8 Fast.
+
 ### Organizing Agents in Your Repository
 
 ```

website/src/content/docs/learning-hub/creating-effective-skills.md

@@ -3,7 +3,7 @@ title: 'Creating Effective Skills'
 description: 'Master the art of writing reusable, shareable skill folders that deliver consistent results across your team.'
 authors:
   - GitHub Copilot Learning Hub Team
-lastUpdated: 2026-06-30
+lastUpdated: 2026-07-05
 estimatedReadingTime: '9 minutes'
 tags:
   - skills
@@ -413,6 +413,17 @@ A: In v1.0.66+, the agent can propose draft skill additions or improvements as i
 
 This opens a review flow where you can accept, reject, or defer each proposed change — giving you full control over how your skill library evolves. No changes are applied until you approve them.
 
+**Q: How does Copilot decide which skills to load from a large library?**
+
+A: By default, Copilot CLI uses embeddings-based retrieval to dynamically select which skills are relevant to the current task (v1.0.66+). You can control this with the `--dynamic-retrieval` flag:
+
+```bash
+copilot --dynamic-retrieval skills=on   # enable dynamic skill retrieval (default)
+copilot --dynamic-retrieval skills=off  # always load all skills into context
+```
+
+You can also set this persistently via the `dynamicRetrieval` key in CLI settings. Dynamic retrieval is useful when you have a large number of skills and want to keep context lean — only the most relevant skills are loaded per turn. Turning it off ensures every skill is always available but uses more context budget.
+
 ## Common Pitfalls to Avoid
 
 - ❌ **Vague description**: "Code helper" doesn't help agents discover the skill

website/src/content/docs/learning-hub/github-copilot-app.md

@@ -3,7 +3,7 @@ title: 'Getting Started with the GitHub Copilot app'
 description: 'Learn about the GitHub Copilot app, a desktop experience built for agent-native development. Understand its key features and who it''s for.'
 authors:
   - GitHub Copilot Learning Hub Team
-lastUpdated: 2026-06-17
+lastUpdated: 2026-07-05
 estimatedReadingTime: '8 minutes'
 tags:
   - copilot-app
@@ -88,6 +88,8 @@ For a hands-on guide to building canvases with `/create-canvas`, see [Working wi
 
 You control the automation level—decide whether Agent Merge should just run CI, address feedback, or go all the way to merging. It's a way to let Copilot handle the tedious parts of the review and merge process.
 
+**`/pr auto` self-paced loop** *(v1.0.66+)*: The `/pr auto` command now runs as a self-paced loop — fixing one issue per run, then automatically pacing itself around CI checks to keep driving the PR toward green. It continues iterating without requiring manual re-triggering. Use `/pr automerge` to go a step further: it keeps going until the PR is fully merged. You can manage or stop either loop at any time from `/loop` or `/every`.
+
 ## Who is the Copilot app for?
 
 The Copilot app isn't a replacement for existing Copilot experiences—it's another tool in the toolbox. Here's who it serves best:
@@ -276,7 +278,8 @@ Enable Agent Merge to automate routine PR workflows:
 2. Specify what automations to enable (run CI, address feedback, merge)
 3. Create a session to implement a feature
 4. When the PR is created, Agent Merge monitors it
-5. It runs CI, waits for reviews, addresses feedback, and merges when ready
+5. `/pr auto` runs a self-paced loop, fixing one issue per run and pacing around CI checks
+6. Use `/pr automerge` to keep going until the PR is merged; manage the loop from `/loop` or `/every`
 
 ## Next Steps