Python: Stop skill discovery below skill boundaries

[Lovlace777]

Motivation & Context

Summary

File-based skill discovery currently keeps descending after it finds a SKILL.md. This can treat content nested beneath a skill boundary as an independent skill root. Stop discovery at the first skill boundary so everything below it remains part of that skill.

Description & Review Guide

• What are the major changes? Return immediately after recording a directory that contains SKILL.md, and add a regression test with a nested SKILL.md.
• What is the impact of these changes? Standard flat skill layouts are unchanged, while nested skill content is no longer discovered as a separate skill.
• What do you want reviewers to focus on? The boundary behavior in _discover_skill_directories and the regression test structure.

Validation

• uv run pytest packages/core/tests/core/test_skills.py::TestDiscoverSkillDirectories -q — 7 passed.
• uv run pytest packages/core/tests/core/test_skills.py -q — passed.
• uv run poe test -P core — 3350 passed, 30 skipped, 2 xfailed.
• uv run poe syntax -P core -C — passed.
• uv run pyright packages/core/agent_framework/_skills.py — 0 errors.
• uv run pyright -p pyrightconfig.tests.json packages/core/tests/core/test_skills.py — 0 errors.
• git diff --check — passed.

Notes

No public API, dependency, or compatibility changes. The full core Pyright sweep could not resolve locally absent optional Graphviz, OpenTelemetry exporter, Azure Agent Server, and Azure Content Understanding dependencies; focused strict checks for both changed files pass.

Related Issue

Fixes #6682

Contribution Checklist

• The code builds clean without any errors or warnings
• All unit tests pass, and I have added new tests where possible
• The PR follows the Contribution Guidelines
• This PR is linked to an issue and there is no other open PR for this issue (see Related Issue above).
• This is not a breaking change. If it is a breaking change, add the breaking change label (or add "[BREAKING]" to the title prefix, before or after any language prefix) — a workflow keeps the label and the title prefix in sync automatically.

[Copilot]

Pull request overview

Stops file-based skill discovery from recursing into subdirectories once a skill boundary (SKILL.md) is found, ensuring nested content remains part of the parent skill rather than being treated as separate skill roots.

Changes:

• Update FileSkillsSource._discover_skill_directories to return immediately after recording a directory containing SKILL.md.
• Add a regression test covering a nested SKILL.md under a parent skill boundary.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
python/packages/core/agent_framework/_skills.py	Adds an early return in directory traversal to stop descending below a directory containing SKILL.md.
python/packages/core/tests/core/test_skills.py	Adds a regression test asserting nested skill roots are not discovered beneath a parent SKILL.md boundary.

[github-actions]

Python Test Coverage Report •

File	Stmts	Miss	Cover	Missing
packages/core/agent_framework
_skills.py	1022	37	96%	294, 541, 1007, 1022, 1024–1025, 1381–1382, 1394–1395, 1625, 1654, 2117, 2573–2574, 2709, 2714, 2717, 2722, 2749, 2754, 2808, 2817, 2822, 2825, 2830, 2854, 2859, 3102–3103, 3452, 3679–3680, 3707–3708, 3715–3716
TOTAL	40697	4626	88%

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
8119	34 💤	0 ❌	0 🔥	2m 12s ⏱️

[SergeyMenshykh]

@Lovlace777 please answer the CLA requirement described above to proceed with the PR.

[itxaiohanglover]

Clean fix! Returning early after finding a SKILL.md prevents descending into nested skill directories. The test with parent-skill containing nested-skill is the perfect regression case.

[giles17]

@Lovlace777 reminder to answer the CLA requirement so we can merge the PR

[giles17]

Closing as fix is no longer valid