Python: Fix read_skill_resource instruction dropping .md extension

[giles17]

Motivation & Context

Microsoft Agent Framework Skills expose a read_skill_resource tool that resolves resources by their exact listed name. For file-based skills, resource names include their relative path and file extension (e.g. references/output-schema.md), and the <available_resources> block advertises them that way. Lookup (get_resource) matches on that exact name.

However, the generated RESOURCE_INSTRUCTIONS prompt contained a misleading example that told the model to strip the extension:

Use read_skill_resource ... (e.g. "style-guide" not "style-guide.md", "references/FAQ" not "FAQ.md").

The second example (references/FAQ instead of references/FAQ.md) demonstrates dropping the .md extension, which directly contradicts the exact-match behavior. This nudges the model into calling read_skill_resource(resource_name="references/output-schema"), which then fails with Resource '...' not found in skill '...'.

This is a Python-only transcription defect: the .NET original (AgentSkillsProvider.cs) correctly reads "references/FAQ.md" not "FAQ.md".

Description & Review Guide

• What are the major changes? A one-line fix to the RESOURCE_INSTRUCTIONS example in python/packages/core/agent_framework/_skills.py: "references/FAQ" -> "references/FAQ.md", restoring exact parity with the .NET prompt. The unifying rule ("use the name exactly as listed") is unchanged; the file-resource example now keeps its extension.
• What is the impact of these changes? Agents following the instruction now call read_skill_resource with the full listed name (including .md), so resource loading no longer fails. Prompt-text only; no API or behavioral code changes.
• What do you want reviewers to focus on? Confirm the corrected example matches how file resources are named/advertised and looked up.

Related Issue

Fixes #7008

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 title prefix in sync automatically.

[Copilot]

Pull request overview

This PR fixes a misleading example in the Python skills system prompt (RESOURCE_INSTRUCTIONS) so that read_skill_resource is called with the exact resource name as listed in <available_resources>, including the .md extension for file-based resources (aligning Python with the .NET prompt).

Changes:

• Update the RESOURCE_INSTRUCTIONS example from references/FAQ to references/FAQ.md to match exact-name lookup behavior.
• Preserve the existing rule text (“use the name exactly as listed”) while correcting the file-resource example that previously contradicted lookup semantics.

[github-actions]

Automated Code Review

Reviewers: 5 | Confidence: 96% | Result: All clear

Reviewed: Correctness, Security Reliability, Test Coverage, Failure Modes, Design Approach

---

Automated review by giles17's agents

[github-actions]

Python Test Coverage Report •

File	Stmts	Miss	Cover	Missing
packages/core/agent_framework
_skills.py	1083	39	96%	314, 584, 1102, 1117, 1119–1120, 1486–1487, 1729, 1758, 2252, 2441, 2949–2950, 3047, 3055, 3060, 3063, 3068, 3088, 3100, 3105, 3200, 3208, 3213, 3216, 3221, 3241, 3250, 3255, 3521–3522, 4074, 4322–4323, 4350–4351, 4358–4359
TOTAL	44098	5270	88%

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
8833	33 💤	0 ❌	0 🔥	2m 10s ⏱️