Enable AMD-AGI/TraceLens to reproduce minimal tests required by amd/skills

.coveragerc

@@ -6,6 +6,7 @@ omit =
     TraceLens/PerfModel/benchmarking/*
     TraceLens/PerfModel/origami_helper.py
     TraceLens/PerfModel/run_perf_model.py
+    TraceLens/Agent/*/skills/*/evals/*
 
 [paths]
 source =

.github/workflows/skill-evals.yml

@@ -0,0 +1,45 @@
+###############################################################################
+# Copyright (c) 2026 Advanced Micro Devices, Inc. All rights reserved.
+#
+# See LICENSE for license information.
+###############################################################################
+
+# Run skillscope on evals.json (same intake criteria as amd/skills)
+
+name: AMD Skills Checks
+
+on:
+  pull_request:
+    paths:
+      - "TraceLens/Agent/Analysis/skills/**"
+      - ".github/workflows/skill-evals.yml"
+  workflow_dispatch:
+    inputs:
+      structural:
+        description: Structural Tests
+        type: boolean
+        default: true
+      routing:
+        description: Routing Tests
+        type: boolean
+        default: false
+      behavioral:
+        description: Behavioral Tests
+        type: boolean
+        default: false
+
+permissions:
+  contents: read
+
+jobs:
+  evals:
+    name: AMD Skills Checks
+    uses: amd/skillscope/.github/workflows/reusable.yml@main
+    secrets:
+      api_key: ${{ secrets.ANTHROPIC_API_KEY }}
+    with:
+      skills: TraceLens/Agent/Analysis/skills/analysis-orchestrator
+      # Each step can be set to `required`, `optional`, or `off`.
+      structural: ${{ github.event_name != 'workflow_dispatch' && 'required' || inputs.structural && 'required' || 'off' }}
+      routing: ${{ github.event_name == 'workflow_dispatch' && inputs.routing && 'required' || 'off' }}
+      behavioral: ${{ github.event_name == 'workflow_dispatch' && inputs.behavioral && 'required' || 'off' }}

TraceLens/Agent/Analysis/skills/analysis-orchestrator/SKILL.md

@@ -1,5 +1,5 @@
 ---
-name: tracelens-analysis-orchestrator
+name: analysis-orchestrator
 description: >-
   Orchestrates modular PyTorch profiler trace analysis with TraceLens: generates perf
   reports, prepares category data, runs system-level and compute-kernel subagents in

TraceLens/Agent/Analysis/skills/analysis-orchestrator/evals/evals.json

@@ -0,0 +1,50 @@
+{
+  "evaluations": [
+    {
+      "id": "gemm-01-repeatability",
+      "skill_should_trigger": true,
+      "prompt": "Run the full standalone TraceLens analysis workflow on locally with the default analysis mode, and write everything to analysis_output/ in the current directory. The trace ships inside https://github.com/AMD-AGI/TraceLens/raw/main/agent_evals/Analysis/analysis_tests/unit_tests_standalone.tar.gz. Unpack that and analyze agent_evals/Analysis/analysis_tests/unit_tests_standalone/gemm/gemm_01_compute_few_tiles_analysis_output/gemm_01_compute_few_tiles.json.",
+      "files_exist": [
+        "analysis_output/analysis.md",
+        "analysis_output/perf_report.xlsx",
+        "analysis_output/category_data/category_manifest.json",
+        "analysis_output/priority_data.json",
+        "analysis_output/metadata/model_info.json",
+        "analysis_output/perf_improvement.png"
+      ],
+      "expected_behavior": [
+        "Filled the Executive Summary of analysis.md with measured numbers, leaving no template placeholders such as 'X ms', 'Y%' or '<Brief Title>' behind"
+      ]
+    },
+    {
+      "id": "analyze-torch-trace",
+      "skill_should_trigger": true,
+      "prompt": "My training step is way slower than I expected. I ran it under the PyTorch profiler and the output is at https://github.com/AMD-AGI/TraceLens/raw/main/tests/traces/mi300/resnet_act_checkpoint.json.gz. Work out where the time is actually going and write it up so I know what to fix first."
+    },
+    {
+      "id": "compare-two-traces",
+      "skill_should_trigger": true,
+      "prompt": "I profiled the same model on two different machines and one is clearly slower, but I can't tell why. The captures are at https://github.com/AMD-AGI/TraceLens/raw/main/tests/traces/mi300/facebook_timesformer-base-finetuned-k400__1016002.json.gz and https://github.com/AMD-AGI/TraceLens/raw/main/tests/traces/h100/facebook_timesformer-base-finetuned-k400__1016002.json.gz. Tell me which GPU kernels are worse on the slow one, worst first, and write it up."
+    },
+    {
+      "id": "agentic-analysis-workflow",
+      "skill_should_trigger": true,
+      "prompt": "Run the agentic analysis workflow on the trace at https://github.com/AMD-AGI/TraceLens/raw/main/tests/traces/mi300/gaunernst_bert-small-uncased__1016001.json.gz and produce analysis.md."
+    },
+    {
+      "id": "kernel-perf-report",
+      "skill_should_trigger": true,
+      "prompt": "I captured a GPU kernel trace from a distributed training run: https://github.com/AMD-AGI/TraceLens/raw/main/tests/traces/mi300/llama_70b_fsdp/rank0_trace_no_pyfn.json.gz. I need a prioritized report of the worst performance offenders that I can show stakeholders."
+    },
+    {
+      "id": "node-event-loop",
+      "skill_should_trigger": false,
+      "prompt": "My Node.js server shows high event loop latency under load. Help me profile it and find the blocking call."
+    },
+    {
+      "id": "cprofile-python-script",
+      "skill_should_trigger": false,
+      "prompt": "Profile my Python script with cProfile and tell me which function is eating the most time."
+    }
+  ]
+}

TraceLens/Agent/Analysis/skills/analysis-orchestrator/evals/machine.yml

@@ -0,0 +1,7 @@
+###############################################################################
+# Copyright (c) 2026 Advanced Micro Devices, Inc. All rights reserved.
+#
+# See LICENSE for license information.
+###############################################################################
+
+os: [Linux]