Add continuous-eval observe ref for Foundry agent Monitoring

[jugonzales]

Description

Adds a new continuous-eval sub-skill that documents the continuous_eval_create, continuous_eval_get, and
continuous_eval_delete MCP tools. These tools enable ongoing evaluation of agent responses — auto-detecting
agent kind and routing to the appropriate backend (evaluation rules for prompt/workflow agents, scheduled
evaluations for hosted agents).

Changes:

- New foundry-agent/continuous-eval/continuous-eval.md — skill doc with entry points, behavioral rules, 

operations, response format, and evaluator guidance
- SKILL.md — registered continuous-eval sub-skill in the sub-skills table and lifecycle table
- observe/observe.md — added cross-reference to continuous-eval in Related Skills
- Updates snapshots to handle keywords

Checklist

• Tests pass locally (cd tests && npm test)
• If modifying skill descriptions: verified routing correctness with integration tests (npm run test:skills:integration -- <skill>)
• If modifying skill USE FOR / DO NOT USE FOR / PREFER OVER clauses: confirmed no routing regressions for competing skills
• Version bumped in skill frontmatter (if skill files changed)

[Copilot]

Pull request overview

Adds documentation and routing keywords for Foundry “continuous evaluation” monitoring, integrating it into the existing observe workflow and updating trigger keyword snapshots accordingly.

Changes:

• Added a new continuous evaluation reference doc covering continuous_eval_create/get/delete and how to act on monitoring results.
• Updated observe step-6 monitoring guidance and expanded observe entry points/keywords to include production monitoring scenarios.
• Bumped microsoft-foundry skill version and refreshed trigger keyword snapshots to reflect the updated description/keywords.

Reviewed changes

Copilot reviewed 17 out of 17 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
plugin/skills/microsoft-foundry/SKILL.md	Updates skill description/keywords, bumps version, and clarifies observe lifecycle entry for continuous monitoring.
plugin/skills/microsoft-foundry/foundry-agent/observe/observe.md	Adds continuous monitoring intents, tools, and entry point to the observe skill.
plugin/skills/microsoft-foundry/foundry-agent/observe/references/cicd-monitoring.md	Refocuses Step 6 on CI/CD eval gates plus continuous production monitoring and links to continuous eval reference.
plugin/skills/microsoft-foundry/foundry-agent/observe/references/continuous-eval.md	New reference describing continuous evaluation configuration, operations, and remediation loop.
tests/microsoft-foundry/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords (continuous/enable/disable, etc.).
tests/microsoft-foundry/resource/create/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords.
tests/microsoft-foundry/models/deploy/deploy-model/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords.
tests/microsoft-foundry/models/deploy/deploy-model-optimal-region/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords.
tests/microsoft-foundry/models/deploy/customize-deployment/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords.
tests/microsoft-foundry/models/deploy/capacity/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords.
tests/microsoft-foundry/foundry-agent/create/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords.
tests/microsoft-foundry/foundry-agent/deploy/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords.
tests/microsoft-foundry/foundry-agent/invoke/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords.
tests/microsoft-foundry/foundry-agent/observe/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords.
tests/microsoft-foundry/foundry-agent/trace/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords.
tests/microsoft-foundry/foundry-agent/troubleshoot/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords.
tests/microsoft-foundry/foundry-agent/eval-datasets/snapshots/triggers.test.ts.snap	Snapshot update for changed skill description keywords.

Comments suppressed due to low confidence (1)

plugin/skills/microsoft-foundry/SKILL.md:28

• PR description says a new doc was added at foundry-agent/continuous-eval/continuous-eval.md, but the change actually adds foundry-agent/observe/references/continuous-eval.md and no new foundry-agent/continuous-eval/ sub-skill entry appears in the Sub-Skills table. Either update the PR description to match, or add/register the intended standalone sub-skill to avoid confusion for maintainers.

| Sub-Skill | When to Use | Reference |
|-----------|-------------|-----------|
| **deploy** | Containerize, build, push to ACR, create/update/start/stop/clone agent deployments | [deploy](foundry-agent/deploy/deploy.md) |
| **invoke** | Send messages to an agent, single or multi-turn conversations | [invoke](foundry-agent/invoke/invoke.md) |
| **observe** | Evaluate agent quality, run batch evals, analyze failures, optimize prompts, improve agent instructions, compare versions, set up CI/CD monitoring, and enable continuous production evaluation | [observe](foundry-agent/observe/observe.md) |
| **trace** | Query traces, analyze latency/failures, correlate eval results to specific responses via App Insights `customEvents` | [trace](foundry-agent/trace/trace.md) |
| **troubleshoot** | View container logs, query telemetry, diagnose failures | [troubleshoot](foundry-agent/troubleshoot/troubleshoot.md) |
| **create** | Create new hosted agent applications. Supports Microsoft Agent Framework, LangGraph, or custom frameworks in Python or C#. Downloads starter samples from foundry-samples repo. | [create](foundry-agent/create/create.md) |
| **eval-datasets** | Harvest production traces into evaluation datasets, manage dataset versions and splits, track evaluation metrics over time, detect regressions, and maintain full lineage from trace to deployment. Use for: create dataset from traces, dataset versioning, evaluation trending, regression detection, dataset comparison, eval lineage. | [eval-datasets](foundry-agent/eval-datasets/eval-datasets.md) |

[jongio]

The new continuous-eval.md is well-structured - clear entry points, behavioral rules, and operations with proper cross-references to trace/deploy/observe. Separating pre-deploy (CI/CD pipeline) from post-deploy (continuous monitoring) in cicd-monitoring.md is a solid design choice. Version bump and snapshot updates look correct.

Two things to address before merging:

1. cicd-monitoring.md duplicates roughly 70% of continuous-eval.md's remediation content (score reading, triage steps, routing table, verification). The two docs already have small wording drift between them. Since cicd-monitoring.md already links to continuous-eval.md for setup, it should also defer to it for the "acting on results" workflow instead of duplicating it inline.

2. observe.md Quick Reference lists continuous_eval_create and continuous_eval_get but not continuous_eval_delete. The linked continuous-eval.md documents delete as a full operation with its own entry point - the parent should surface it too.

[jongio]

This entire section (reading scores, triage, remediation routing, verification) is nearly identical to continuous-eval.md's "Acting on Results" - already with small wording drift (e.g., "and timestamps" here but not there, "Route To" vs "Action" column names). Since you're already linking to continuous-eval.md for the setup workflow, consider replacing this with:
For how to read evaluation scores, triage regressions, and verify fixes, see Acting on Results.
That keeps continuous-eval.md as the single source of truth for the remediation loop.

[jongio]

continuous_eval_delete is missing from this list but is documented as a full operation in the linked continuous-eval.md reference (with its own entry point for "Delete continuous eval"). Consider adding it here for consistency.

[Copilot]

Pull request overview

Copilot reviewed 17 out of 17 changed files in this pull request and generated 4 comments.

[jongio]

The new continuous-eval.md is solid - entry points, behavioral rules, and operations are well-structured with proper cross-references. Four additional items beyond what's already flagged:

1. The Disable operation likely destroys evaluator config if the tool upserts (see inline comment).
2. evaluation_get isn't in the Quick Reference but is used in the remediation workflow.
3. The observe.md DO NOT manually call guardrail doesn't cover continuous_eval_create.
4. Evaluator examples differ across three files with no explanation of why.

[jongio]

Three additional items beyond what's already flagged:

1. SKILL.md description dropped the "prompt optimizer workflows" keyword that was in the previous version. The new wording has "prompt optimizer" but not "prompt optimizer workflows" - this could regress routing for that phrase. Either restore it or confirm the removal was intentional.

2. The scenario parameter in continuous-eval.md's optional parameters table lists standard and business values but doesn't explain what each mode does or when to choose one. An agent can't make a useful recommendation without this context.

3. continuous_eval_get returns a list (per the Response Format section), but the Disable and Delete workflows assume a single config. If multiple configs exist, there's no guidance on which to target - worth a note on expected cardinality or how to disambiguate.