Use GPT-5.6 Sol and raise agentic workflow label limits

[JeffreyCA]

Fixes #9975

Summary

This PR configures all three agentic workflows to use GPT-5.6 Sol with medium reasoning effort. Previously, the workflows relied on Copilot's implicit auto model selection. Observed runs resolved that to Claude Sonnet models, including claude-sonnet-4.6 and claude-sonnet-5. The September 7 instruction-update run failed because Sonnet 4.6 was no longer available to the agentic-workflows integrator.

Explicitly selecting GPT-5.6 Sol avoids that model drift and uses the preferred OpenAI model for cost and performance.

Label limits

The safe-output label limits are raised from 5 to 10 for issue triage and from 10 to 25 for extension PR labelling. The extension limit now covers bulk changes that touch many extension folders.

Model configuration

Each workflow selects copilot/gpt-5.6-sol explicitly. Reasoning effort is passed through engine.args as --effort medium because Copilot CLI treats gpt-5.6-sol?effort=medium as a literal model ID and rejects it as unsupported.

The generated workflows and action lock were refreshed with gh-aw v0.88.7.

Testing

• Compiled and validated all three workflows with gh-aw v0.88.7.
• Deployed all three generated workflows to a fork and confirmed GPT-5.6 Sol ran successfully with medium reasoning effort.
• Confirmed issue triage applied the expected issue labels.
• Confirmed the extension PR labeler completed end to end and applied ext-agents to docs(ai agents): clarify --no-inspector usage JeffreyCA/azure-dev#102.
• Dispatched the instruction-update workflow over a recent 10-PR range with high promotion thresholds. Every job passed and the no-change run created no PR or issue.

[azure-pipelines]

Azure Pipelines:
22 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[azure-pipelines]

Azure Pipelines:
22 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[Copilot]

Copilot review overview

🔵 Needs a closer look

The regenerated pull_request_target workflow substantially changes security-sensitive runtime, telemetry, container, and threat-detection plumbing beyond the targeted configuration updates.

Review tier: Balanced
Findings: None

What changed in this PR

Configures agentic triage workflows to use GPT-5.6 Sol with medium reasoning and increases label limits.

Changes:

• Pins both workflows to copilot/gpt-5.6-sol.
• Raises issue and extension PR label limits to 10 and 25.
• Refreshes generated workflow dependencies with gh-aw v0.88.7.

File	Description
.github/​workflows/​issue-triage.md	Configures the model, effort, and issue-label limit.
.github/​workflows/​extension-pr-labeler.md	Configures the model, effort, and PR-label limit.
.github/​workflows/​extension-pr-labeler.lock.yml	Regenerates the executable workflow and dependency pins.
.github/​aw/​actions-lock.json	Updates the gh-aw setup action pin.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[azure-pipelines]

Azure Pipelines:
22 pipeline(s) were filtered out due to trigger conditions.
There may be pipelines that require an authorized user to comment /azp run to run.

[Copilot]

Copilot review overview

🟢 Approval recommended

The source and generated workflows are consistent, satisfy #9975, and pass the relevant validation checks.

Review tier: Balanced
Findings: None

[JeffreyCA]

/check-enforcer override