diff --git a/.agents/skills/azure-foundry-finetuning b/.agents/skills/azure-foundry-finetuning
new file mode 120000
index 0000000..4ab200a
--- /dev/null
+++ b/.agents/skills/azure-foundry-finetuning
@@ -0,0 +1 @@
+../../.github/skills/azure-foundry-finetuning
\ No newline at end of file
diff --git a/.claude/skills/azure-foundry-finetuning b/.claude/skills/azure-foundry-finetuning
new file mode 120000
index 0000000..4ab200a
--- /dev/null
+++ b/.claude/skills/azure-foundry-finetuning
@@ -0,0 +1 @@
+../../.github/skills/azure-foundry-finetuning
\ No newline at end of file
diff --git a/.github/skills/azure-foundry-finetuning/SKILL.md b/.github/skills/azure-foundry-finetuning/SKILL.md
new file mode 100644
index 0000000..941c963
--- /dev/null
+++ b/.github/skills/azure-foundry-finetuning/SKILL.md
@@ -0,0 +1,31 @@
+---
+name: azure-foundry-finetuning-py
+description: Submit fine-tuning jobs on Azure AI Foundry.
+compatibility: Requires uv python package
+license: MIT
+metadata:
+  author: Microsoft
+  version: "1.0.0"
+  package: azure-ai-projects
+---
+
+# Azure AI Foundry Fine-Tuning Py
+
+## When to use this
+
+This skill helps users submit supervised fine-tuning (SFT) jobs on Azure AI Foundry. Confirm with the user which arguments are used for the fine-tuning job before submission.
+
+## Prerequisite enforcement
+
+Before running any script in this skill, you must verify `uv` is available:
+
+`command -v uv >/dev/null 2>&1 || pip install uv`
+
+If `uv` is still unavailable after installation, stop and report the prerequisite failure. Do not run this skill's scripts without `uv`.
+
+## Available scripts
+
+Use --help flag to see the usage of each script, e.g. `uv run scripts/submit_sft.py --help`.
+
+- **`scripts/submit_sft.py`** — Submits SFT fine-tuning jobs to Azure AI Foundry. After successfully submitting the job, ask user if they want to monitor the job, and if yes, call `monitor_ft_job.py` with the returned job ID.
+- **`scripts/monitor_ft_job.py`** — Monitors a fine-tuning job by job ID. Stream the output of monitoring to the user until the job is completed, and report the final status of the job.
\ No newline at end of file
diff --git a/.github/skills/azure-foundry-finetuning/scripts/common.py b/.github/skills/azure-foundry-finetuning/scripts/common.py
new file mode 100644
index 0000000..c7b397b
--- /dev/null
+++ b/.github/skills/azure-foundry-finetuning/scripts/common.py
@@ -0,0 +1,10 @@
+import argparse
+import sys
+
+
+class HelpOnErrorParser(argparse.ArgumentParser):
+    """ArgumentParser that prints full help when parsing fails."""
+
+    def error(self, message):
+        self.print_help(sys.stderr)
+        self.exit(2, f"\nerror: {message}\n")
\ No newline at end of file
diff --git a/.github/skills/azure-foundry-finetuning/scripts/monitor_ft_job.py b/.github/skills/azure-foundry-finetuning/scripts/monitor_ft_job.py
new file mode 100644
index 0000000..f17ee89
--- /dev/null
+++ b/.github/skills/azure-foundry-finetuning/scripts/monitor_ft_job.py
@@ -0,0 +1,71 @@
+#!/usr/bin/env python3
+# /// script
+# dependencies = [
+#   "azure-identity",
+#   "azure-ai-projects",
+# ]
+# ///
+import argparse
+import sys
+import time
+from azure.identity import DefaultAzureCredential
+from azure.ai.projects import AIProjectClient
+from common import HelpOnErrorParser
+
+
+TERMINAL_STATUSES = {"succeeded", "failed", "cancelled"}
+
+
+def monitor_ft_job(args):
+	credential = DefaultAzureCredential()
+	project_client = AIProjectClient(endpoint=args.project_endpoint, credential=credential)
+	openai_client = project_client.get_openai_client()
+
+	print(f"Monitoring fine-tuning job: {args.job_id}")
+	while True:
+		job = openai_client.fine_tuning.jobs.retrieve(args.job_id)
+		status = (getattr(job, "status", "") or "").lower()
+		print(f"Job status: {status}")
+
+		if status in TERMINAL_STATUSES:
+			print(f"Job reached terminal status: {status}")
+			if status == "succeeded":
+				return
+			raise RuntimeError(f"Fine-tuning job ended with terminal status: {status}")
+
+		time.sleep(args.poll_interval)
+
+
+def build_parser():
+	parser = HelpOnErrorParser(
+		description="Monitor an SFT fine-tuning job until completion",
+		epilog=(
+			"Example:\n"
+			"  ./monitor_ft_job.py --project-endpoint https://<resource>.services.ai.azure.com/api/projects/<project> --job-id ftjob_123"
+		),
+		formatter_class=argparse.RawTextHelpFormatter,
+	)
+	parser.add_argument(
+		"--project-endpoint",
+		required=True,
+		help="Azure AI Project endpoint URL",
+	)
+	parser.add_argument("--job-id", required=True, help="Fine-tuning job ID")
+	parser.add_argument(
+		"--poll-interval",
+		type=int,
+		default=10,
+		help="Polling interval in seconds (default: 10)",
+	)
+	return parser
+
+
+if __name__ == "__main__":
+	parser = build_parser()
+
+	# If no arguments are provided, show help instead of failing with usage only.
+	if len(sys.argv) == 1:
+		parser.print_help()
+		parser.exit(0)
+
+	monitor_ft_job(parser.parse_args())
diff --git a/.github/skills/azure-foundry-finetuning/scripts/submit_sft.py b/.github/skills/azure-foundry-finetuning/scripts/submit_sft.py
new file mode 100644
index 0000000..dce9d3c
--- /dev/null
+++ b/.github/skills/azure-foundry-finetuning/scripts/submit_sft.py
@@ -0,0 +1,127 @@
+#!/usr/bin/env python3
+# /// script
+# dependencies = [
+#   "azure-identity",
+#   "azure-ai-projects",
+# ]
+# ///
+import argparse
+import sys
+import time
+from azure.identity import DefaultAzureCredential
+from azure.ai.projects import AIProjectClient
+from common import HelpOnErrorParser
+
+
+def upload_and_wait_for_processing(openai_client, path, label, timeout_seconds=600, poll_interval_seconds=5):
+    """Upload a file and wait until it is processed before proceeding."""
+    with open(path, "rb") as file_handle:
+        uploaded = openai_client.files.create(file=file_handle, purpose="fine-tune")
+
+    file_id = getattr(uploaded, "id", None)
+    if not file_id:
+        raise RuntimeError(f"{label} upload did not return a file id.")
+
+    print(f"Uploaded {label} file: {file_id}. Waiting for processing...")
+
+    deadline = time.time() + timeout_seconds
+    while time.time() < deadline:
+        current = openai_client.files.retrieve(file_id)
+        status = (getattr(current, "status", "") or "").lower()
+
+        if status == "processed":
+            print(f"{label} file processed successfully: {file_id}")
+            return file_id
+
+        if status in {"failed", "error", "cancelled"}:
+            raise RuntimeError(
+                f"{label} file processing failed for {file_id} with status '{status}'."
+            )
+
+        time.sleep(poll_interval_seconds)
+
+    raise TimeoutError(
+        f"Timed out after {timeout_seconds}s waiting for {label} file {file_id} to process."
+    )
+
+
+def submit_sft(args):
+    credential = DefaultAzureCredential()
+
+    project_endpoint = args.project_endpoint
+    project_client = AIProjectClient(endpoint=project_endpoint, credential=credential)
+
+    openai_client = project_client.get_openai_client() 
+
+    train_file_id = upload_and_wait_for_processing(
+        openai_client,
+        args.training_file,
+        "training",
+    )
+    val_file_id = upload_and_wait_for_processing(
+        openai_client,
+        args.validation_file,
+        "validation",
+    )
+
+    hyperparameters = {}
+    if args.epochs is not None:
+        hyperparameters["n_epochs"] = args.epochs
+    if args.batch_size is not None:
+        hyperparameters["batch_size"] = args.batch_size
+    if args.learning_rate is not None:
+        hyperparameters["learning_rate_multiplier"] = args.learning_rate
+
+    supervised = {}
+    if hyperparameters:
+        supervised["hyperparameters"] = hyperparameters
+
+    job_kwargs = dict(
+        model=args.model,
+        training_file=train_file_id,
+        validation_file=val_file_id,
+        method={
+            "type": "supervised",
+            "supervised": supervised,
+        },
+        suffix=args.suffix,
+    )
+
+    job = openai_client.fine_tuning.jobs.create(**job_kwargs)
+
+    print("Submitted finetuning job:", job.id)
+
+
+def build_parser():
+    parser = HelpOnErrorParser(
+        description="Submit SFT fine-tuning job",
+        epilog=(
+            "Example:\n"
+            "  ./submit_sft.py --project-endpoint https://<resource>.services.ai.azure.com/api/projects/<project> --training-file train.jsonl --validation-file valid.jsonl --model gpt-4o-mini-2024-07-18"
+        ),
+        formatter_class=argparse.RawTextHelpFormatter,
+    )
+    parser.add_argument(
+        "--project-endpoint",
+        required=True,
+        help="Azure AI Project endpoint URL",
+    )
+    parser.add_argument("--training-file", required=True, help="Path to training JSONL file")
+    parser.add_argument("--validation-file", required=True, help="Path to validation JSONL file")
+    parser.add_argument("--model", required=True, help="Base model name")
+    parser.add_argument("--suffix", default="sft-finetuned", help="Model name suffix")
+    parser.add_argument("--epochs", type=int, default=None, help="Number of epochs")
+    parser.add_argument("--batch-size", type=int, default=None, help="Batch size")
+    parser.add_argument("--learning-rate", type=float, default=None, help="Learning rate multiplier")
+    return parser
+
+
+if __name__ == "__main__":
+    parser = build_parser()
+
+    # If no arguments are provided, show help instead of submitting with defaults.
+    if len(sys.argv) == 1:
+        parser.print_help()
+        parser.exit(0)
+
+    submit_sft(parser.parse_args())
diff --git a/README.md b/README.md
index 292997a..ad2e0cd 100644
--- a/README.md
+++ b/README.md
@@ -7,6 +7,7 @@ This repository contains **12 end-to-end demos** and **sample datasets** for fin
 - [Quick Start](#-quick-start)
 - [Demos](#-demos)
 - [Sample Datasets](#-sample-datasets)
+- [AI Agent Skills](#-ai-agent-skills)
 - [Prerequisites](#-prerequisites)
 - [Contributing](#contributing)
 
@@ -64,6 +65,71 @@ Ready-to-use datasets for testing fine-tuning techniques in the **[Sample_Datase
 
 > ⚠️ **Note**: These datasets are for **learning and experimentation only**—not for production use. Training jobs may incur costs on your Azure subscription.
 
+---
+
+## 🤖 AI Agent Skills
+
+This repo includes the same fine-tuning skill content for multiple coding agents.
+
+- **GitHub Copilot** skills: **[.github/skills](.github/skills/)**
+- **Claude Code** skills: **[.claude/skills](.claude/skills/)**
+- **Codex-compatible agents** skills: **[.agents/skills](.agents/skills/)**
+
+The goal is to keep skill guidance **agent-agnostic** so you can use the same workflows (for example, submitting and monitoring fine-tuning jobs) regardless of the assistant.
+
+### How to use with GitHub Copilot + VS Code
+
+1. Open this repository in **VS Code** with the **GitHub Copilot Chat** extension enabled.
+2. Ask a task in Copilot Chat from the repo context (for example: "help me submit an SFT job with my dataset").
+3. Copilot will match relevant skills from **[.github/skills](.github/skills/)** when your request aligns with a skill description.
+4. If available in your version, invoke a skill slash command directly in chat.
+
+### How to use with Copilot CLI
+
+Use `copilot` from the repository root so it can discover workspace skills in **[.github/skills](.github/skills/)**.
+
+1. Verify the CLI is installed:
+
+	```bash
+	copilot --version
+	```
+
+2. Start a chat session from this repo root:
+
+	```bash
+	cd /path/to/root_of_repo
+	copilot
+	```
+
+3. Ask for the same type of task you would ask in chat, for example:
+
+	```text
+	Help me submit an SFT job with my dataset.
+	```
+
+4. If your CLI version supports slash commands, invoke the skill directly:
+
+	```text
+	/azure-foundry-finetuning
+	```
+
+5. Follow prompts for Azure project details, model, training file, and validation file, then run the suggested scripts.
+
+### How to use with Claude
+
+1. Open this repository in your Claude coding environment from the repo root.
+2. Ask for a fine-tuning workflow task in natural language.
+3. Claude can use skills from **[.claude/skills](.claude/skills/)** when your prompt matches the skill scope.
+4. Follow the generated steps/scripts and provide required Azure inputs when prompted.
+
+### How to use with Codex
+
+1. Open this repository in a Codex-compatible coding agent environment from the repo root.
+2. Ask for the task you want to perform (for example, setting up SFT submission or monitoring a job).
+3. The agent can apply skills from **[.agents/skills](.agents/skills/)** when the request matches a skill.
+4. Run the suggested commands and scripts with your Azure project configuration.
+
+
 ---
 
 ## ✅ Prerequisites