evalrun 0.4.1__tar.gz → 0.4.2__tar.gz

evalrun-0.4.2/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Shivam Bhardwaj
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

{evalrun-0.4.1/evalrun.egg-info → evalrun-0.4.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: evalrun
-Version: 0.4.1
+Version: 0.4.2
 Summary: Provider-agnostic evaluation and regression-testing toolkit for tool-using AI agents.
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown
@@ -10,8 +10,16 @@ Requires-Dist: requests>=2.28.0
 Requires-Dist: pydantic>=2.0.0
 Requires-Dist: python-frontmatter>=1.0.0
 Requires-Dist: PyYAML>=6.0
-Requires-Dist: langfuse>=2.0.0
-Requires-Dist: fastmcp>=0.1.0
+Requires-Dist: markdown-it-py>=3.0.0
+Provides-Extra: observability
+Requires-Dist: langfuse>=2.0.0; extra == "observability"
+Provides-Extra: mcp
+Requires-Dist: fastmcp>=0.1.0; extra == "mcp"
+Requires-Dist: mcp<2,>=1.27; extra == "mcp"
+Provides-Extra: all
+Requires-Dist: langfuse>=2.0.0; extra == "all"
+Requires-Dist: fastmcp>=0.1.0; extra == "all"
+Requires-Dist: mcp<2,>=1.27; extra == "all"
 Dynamic: license-file
 
 # evalrun
@@ -60,6 +68,13 @@ cd agent-eval-platform
 pip install -e .
 ```
 
+Optional integrations are deliberately not required for the core CLI and SDK:
+
+```bash
+pip install "evalrun[observability]"  # optional Langfuse tracing
+pip install "evalrun[mcp]"            # optional MCP validation tools
+```
+
 If your shell says `evalrun: command not found`, activate the virtual environment and install the repository first:
 
 ```bash
@@ -77,7 +92,38 @@ evalrun demo
 open results/demo/report.html  # macOS
 ```
 
-### 3. Single Scenario Run (Hosted Model)
+The offline demo is the safest way to confirm that installation works. It makes
+no model requests, needs no API key, and executes only EvalRun's built-in demo
+agent.
+
+### 3. Create and check your own evaluation workspace
+
+Create starter files in a new folder:
+
+```bash
+evalrun init my-evaluation
+cd my-evaluation
+evalrun validate scenario.md
+```
+
+Edit `scenario.md` to describe the user request, hard constraints, expected
+behavior, scoring dimensions, pass criteria, and failure conditions. The
+scenario is ordinary Markdown; you do not need to change EvalRun's source code.
+Use `evalrun validate` before spending money on a model run. It reports the
+exact missing heading or unsupported dimension and returns exit code `2` when
+the file needs fixing.
+
+Run environment diagnostics at any time with:
+
+```bash
+evalrun doctor
+```
+
+`doctor` does not require an API key. Missing keys and unavailable model servers
+are reported as information, while invalid Python or unwritable output paths are
+reported as failures.
+
+### 4. Single Scenario Run (Hosted Model)
 
 ```bash
 export OPENAI_API_KEY="sk-proj-..."
@@ -103,6 +149,21 @@ EvalRun does not lock you to one model provider. The `--base-url` value is the a
 
 The judge normally uses the same endpoint and API key as the target model. You only need `--judge-base-url` or `--judge-api-key` when the judge is hosted somewhere different. The endpoint URL is not a credential; it simply tells EvalRun where to send the request.
 
+For a target and judge at different providers, configure both explicitly:
+
+```bash
+evalrun run \
+  --scenario path/to/scenario.md \
+  --agent my_agent:MyAgent \
+  --model gemini-3.7-flash \
+  --base-url https://generativelanguage.googleapis.com/v1beta/openai/ \
+  --api-key "$GEMINI_API_KEY" \
+  --judge-model stealth/ox-alpha \
+  --judge-base-url https://openrouter.ai/api/v1 \
+  --judge-api-key "$OPENROUTER_API_KEY" \
+  --output results/mixed-provider-run
+```
+
 For example, a Gemini run can be written as:
 
 ```bash
@@ -118,7 +179,7 @@ evalrun run \
   --output results/gemini-run
 ```
 
-### 4. Local Model Server Run (vLLM / Ollama)
+### 5. Local Model Server Run (vLLM / Ollama)
 
 ```bash
 # Users host their own local OpenAI-compatible server at http://localhost:8000/v1
@@ -132,7 +193,7 @@ evalrun run \
   --output results/local-run
 ```
 
-### 5. Guided Local Web UI
+### 6. Guided Local Web UI
 
 Launch the zero-dependency local web interface:
 
@@ -142,6 +203,17 @@ evalrun ui --port 8501
 
 Open `http://127.0.0.1:8501` in your browser to configure endpoints, select scenarios, run evaluations, view pass/fail/block verdicts, and launch interactive HTML reports.
 
+The UI is a local convenience layer. It supports built-in agents and workspace
+scenario paths. API keys are held in browser memory for the current request and
+are not saved by EvalRun. Do not expose this server to the public internet and
+do not paste untrusted Python code into an agent field. Complex or third-party
+agents should be run through the CLI or SDK in the environment where their code
+and dependencies are installed.
+
+The UI intentionally permits only the built-in travel agent. Custom Python,
+HTTP, and CLI agent adapters remain CLI/SDK-only because accepting arbitrary
+agent specifications from a browser would allow local code execution.
+
 ---
 
 ## 🐍 Python SDK Usage
@@ -164,7 +236,7 @@ results = evaluate(
 report = compare(
     candidate_results=results,
     baseline="results/run-001",
-    max_overall_drop=5.0,
+    max_regression=5.0,
 )
 
 if report.release_blocked:
@@ -274,6 +346,48 @@ evalrun run \
 
 Or place multiple `.md` files in a folder and use `--suite path/to/folder`.
 
+Validate a suite before running it:
+
+```bash
+for scenario in evals/scenarios/my-domain/*.md; do
+  evalrun validate "$scenario" || exit 2
+done
+```
+
+## Connect your own agent
+
+EvalRun does not require agents to use a particular framework. The `--agent`
+value selects one of three adapters:
+
+| Agent form | Example | When to use |
+| --- | --- | --- |
+| Python import | `my_agent:MyAgent` | Your agent is installed/importable in the current environment. |
+| HTTP endpoint | `http://127.0.0.1:8080/predict` | Your agent is running as a local service. |
+| CLI process | `cli:python my_agent.py` | Your agent reads the scenario request from stdin and writes its answer to stdout. |
+
+Python agents receive the target model adapter from EvalRun. HTTP and CLI agents
+must expose the adapter contract described in
+[`docs/architecture.md`](docs/architecture.md). Run `evalrun --help` for the
+complete syntax and examples.
+
+For repeatable runs, put the same values in a JSON or TOML configuration file:
+
+```bash
+evalrun run --config evalrun.json
+```
+
+Command-line flags override values from the configuration file. Keep API keys
+in environment variables and reference them from your shell; never commit them
+to the config file.
+
+## Troubleshooting
+
+- `evalrun: command not found`: activate the virtual environment and run `python -m pip install evalrun` (or `pip install -e .` from a checkout), then run `rehash` in zsh.
+- `Missing required section`: run `evalrun validate path/to/scenario.md` and add the named Markdown heading.
+- Model authentication or endpoint errors: run `evalrun doctor`, confirm the provider's base URL, and check the matching environment variable.
+- A high evaluator score with `BLOCK`: inspect the independent-auditor findings in `report.html`; a quality score is not a release decision.
+- `report.html` is not created after an error: inspect the terminal error and `manifest.json`; failed executions are recorded as failed results rather than silently treated as a passing empty suite.
+
 ## 🏗️ Architecture & Documentation
 
 For detailed system component diagrams, dual-pass auditor sequence flows, and release gate decision trees, see [`docs/architecture.md`](file:///Users/shivam/Projects/AI/agent-eval-platform/docs/architecture.md).

evalrun-0.4.1/PKG-INFO → evalrun-0.4.2/README.md

@@ -1,19 +1,3 @@
-Metadata-Version: 2.4
-Name: evalrun
-Version: 0.4.1
-Summary: Provider-agnostic evaluation and regression-testing toolkit for tool-using AI agents.
-Requires-Python: >=3.10
-Description-Content-Type: text/markdown
-License-File: LICENSE
-Requires-Dist: openai>=1.0.0
-Requires-Dist: requests>=2.28.0
-Requires-Dist: pydantic>=2.0.0
-Requires-Dist: python-frontmatter>=1.0.0
-Requires-Dist: PyYAML>=6.0
-Requires-Dist: langfuse>=2.0.0
-Requires-Dist: fastmcp>=0.1.0
-Dynamic: license-file
-
 # evalrun
 
 > **Quality Score $\neq$ Release Decision.** An LLM evaluator can rate an itinerary 95/100 while an independent auditor blocks it for hard financial violations.
@@ -60,6 +44,13 @@ cd agent-eval-platform
 pip install -e .
 ```
 
+Optional integrations are deliberately not required for the core CLI and SDK:
+
+```bash
+pip install "evalrun[observability]"  # optional Langfuse tracing
+pip install "evalrun[mcp]"            # optional MCP validation tools
+```
+
 If your shell says `evalrun: command not found`, activate the virtual environment and install the repository first:
 
 ```bash
@@ -77,7 +68,38 @@ evalrun demo
 open results/demo/report.html  # macOS
 ```
 
-### 3. Single Scenario Run (Hosted Model)
+The offline demo is the safest way to confirm that installation works. It makes
+no model requests, needs no API key, and executes only EvalRun's built-in demo
+agent.
+
+### 3. Create and check your own evaluation workspace
+
+Create starter files in a new folder:
+
+```bash
+evalrun init my-evaluation
+cd my-evaluation
+evalrun validate scenario.md
+```
+
+Edit `scenario.md` to describe the user request, hard constraints, expected
+behavior, scoring dimensions, pass criteria, and failure conditions. The
+scenario is ordinary Markdown; you do not need to change EvalRun's source code.
+Use `evalrun validate` before spending money on a model run. It reports the
+exact missing heading or unsupported dimension and returns exit code `2` when
+the file needs fixing.
+
+Run environment diagnostics at any time with:
+
+```bash
+evalrun doctor
+```
+
+`doctor` does not require an API key. Missing keys and unavailable model servers
+are reported as information, while invalid Python or unwritable output paths are
+reported as failures.
+
+### 4. Single Scenario Run (Hosted Model)
 
 ```bash
 export OPENAI_API_KEY="sk-proj-..."
@@ -103,6 +125,21 @@ EvalRun does not lock you to one model provider. The `--base-url` value is the a
 
 The judge normally uses the same endpoint and API key as the target model. You only need `--judge-base-url` or `--judge-api-key` when the judge is hosted somewhere different. The endpoint URL is not a credential; it simply tells EvalRun where to send the request.
 
+For a target and judge at different providers, configure both explicitly:
+
+```bash
+evalrun run \
+  --scenario path/to/scenario.md \
+  --agent my_agent:MyAgent \
+  --model gemini-3.7-flash \
+  --base-url https://generativelanguage.googleapis.com/v1beta/openai/ \
+  --api-key "$GEMINI_API_KEY" \
+  --judge-model stealth/ox-alpha \
+  --judge-base-url https://openrouter.ai/api/v1 \
+  --judge-api-key "$OPENROUTER_API_KEY" \
+  --output results/mixed-provider-run
+```
+
 For example, a Gemini run can be written as:
 
 ```bash
@@ -118,7 +155,7 @@ evalrun run \
   --output results/gemini-run
 ```
 
-### 4. Local Model Server Run (vLLM / Ollama)
+### 5. Local Model Server Run (vLLM / Ollama)
 
 ```bash
 # Users host their own local OpenAI-compatible server at http://localhost:8000/v1
@@ -132,7 +169,7 @@ evalrun run \
   --output results/local-run
 ```
 
-### 5. Guided Local Web UI
+### 6. Guided Local Web UI
 
 Launch the zero-dependency local web interface:
 
@@ -142,6 +179,17 @@ evalrun ui --port 8501
 
 Open `http://127.0.0.1:8501` in your browser to configure endpoints, select scenarios, run evaluations, view pass/fail/block verdicts, and launch interactive HTML reports.
 
+The UI is a local convenience layer. It supports built-in agents and workspace
+scenario paths. API keys are held in browser memory for the current request and
+are not saved by EvalRun. Do not expose this server to the public internet and
+do not paste untrusted Python code into an agent field. Complex or third-party
+agents should be run through the CLI or SDK in the environment where their code
+and dependencies are installed.
+
+The UI intentionally permits only the built-in travel agent. Custom Python,
+HTTP, and CLI agent adapters remain CLI/SDK-only because accepting arbitrary
+agent specifications from a browser would allow local code execution.
+
 ---
 
 ## 🐍 Python SDK Usage
@@ -164,7 +212,7 @@ results = evaluate(
 report = compare(
     candidate_results=results,
     baseline="results/run-001",
-    max_overall_drop=5.0,
+    max_regression=5.0,
 )
 
 if report.release_blocked:
@@ -274,6 +322,48 @@ evalrun run \
 
 Or place multiple `.md` files in a folder and use `--suite path/to/folder`.
 
+Validate a suite before running it:
+
+```bash
+for scenario in evals/scenarios/my-domain/*.md; do
+  evalrun validate "$scenario" || exit 2
+done
+```
+
+## Connect your own agent
+
+EvalRun does not require agents to use a particular framework. The `--agent`
+value selects one of three adapters:
+
+| Agent form | Example | When to use |
+| --- | --- | --- |
+| Python import | `my_agent:MyAgent` | Your agent is installed/importable in the current environment. |
+| HTTP endpoint | `http://127.0.0.1:8080/predict` | Your agent is running as a local service. |
+| CLI process | `cli:python my_agent.py` | Your agent reads the scenario request from stdin and writes its answer to stdout. |
+
+Python agents receive the target model adapter from EvalRun. HTTP and CLI agents
+must expose the adapter contract described in
+[`docs/architecture.md`](docs/architecture.md). Run `evalrun --help` for the
+complete syntax and examples.
+
+For repeatable runs, put the same values in a JSON or TOML configuration file:
+
+```bash
+evalrun run --config evalrun.json
+```
+
+Command-line flags override values from the configuration file. Keep API keys
+in environment variables and reference them from your shell; never commit them
+to the config file.
+
+## Troubleshooting
+
+- `evalrun: command not found`: activate the virtual environment and run `python -m pip install evalrun` (or `pip install -e .` from a checkout), then run `rehash` in zsh.
+- `Missing required section`: run `evalrun validate path/to/scenario.md` and add the named Markdown heading.
+- Model authentication or endpoint errors: run `evalrun doctor`, confirm the provider's base URL, and check the matching environment variable.
+- A high evaluator score with `BLOCK`: inspect the independent-auditor findings in `report.html`; a quality score is not a release decision.
+- `report.html` is not created after an error: inspect the terminal error and `manifest.json`; failed executions are recorded as failed results rather than silently treated as a passing empty suite.
+
 ## 🏗️ Architecture & Documentation
 
 For detailed system component diagrams, dual-pass auditor sequence flows, and release gate decision trees, see [`docs/architecture.md`](file:///Users/shivam/Projects/AI/agent-eval-platform/docs/architecture.md).

{evalrun-0.4.1 → evalrun-0.4.2}/agents/reflection/agent.py

@@ -1,10 +1,10 @@
 """Reflection agent implementation for auditing and critiquing planned itineraries."""
 
 from typing import Optional
-from langfuse import observe
 from framework.llms import BaseLLM, Message
 from framework.models import AgentOutput
 from framework.memory import BaseSessionMemory
+from framework.observability import observe
 from agents.base import BaseAgent
 from .prompts import REFLECTION_SYSTEM_PROMPT
 

{evalrun-0.4.1 → evalrun-0.4.2}/agents/research/agent.py

@@ -1,7 +1,6 @@
-from langfuse import observe
-
 from framework.llms import BaseLLM, Message
 from framework.models import AgentOutput
+from framework.observability import observe
 from agents.base import BaseAgent
 from .prompts import RESEARCH_SYSTEM_PROMPT
 

{evalrun-0.4.1 → evalrun-0.4.2}/agents/research/planner.py

@@ -2,8 +2,7 @@
 
 from framework.llms import BaseLLM, Message
 from framework.utils import parse_json_markdown
-
-from langfuse import observe
+from framework.observability import observe
 
 
 class ResearchPlanner:

{evalrun-0.4.1 → evalrun-0.4.2}/agents/support/triage_agent.py

@@ -1,10 +1,9 @@
 """Customer-support triage agent used by the second-domain benchmark."""
 
-from langfuse import observe
-
 from agents.base import BaseAgent
 from framework.llms import BaseLLM, Message
 from framework.models import AgentOutput
+from framework.observability import observe
 
 
 SUPPORT_TRIAGE_SYSTEM_PROMPT = """You are a production customer-support triage agent.

{evalrun-0.4.1 → evalrun-0.4.2}/agents/travel/agent.py

@@ -1,5 +1,3 @@
-from langfuse import observe
-
 import asyncio
 import json
 from typing import Any, Dict, Optional
@@ -7,6 +5,7 @@ from framework.llms import BaseLLM, Message
 from framework.models import AgentOutput
 from framework.utils import parse_json_markdown
 from framework.memory import BaseSessionMemory
+from framework.observability import observe
 from agents.base import BaseAgent
 from agents.research.planner import ResearchPlanner
 from framework.mcp.revision_summary import (

evalrun-0.4.2/cli/__init__.py

@@ -0,0 +1,6 @@
+"""evalrun CLI package exports."""
+
+from cli.main import create_parser
+from cli.resolver import resolve_agent
+
+__all__ = ["create_parser", "resolve_agent"]

evalrun-0.4.2/cli/doctor.py

@@ -0,0 +1,150 @@
+"""Diagnostic doctor module for evalrun doctor command."""
+
+import importlib
+import os
+import sys
+import urllib.error
+import urllib.request
+from pathlib import Path
+from typing import Dict, List, Tuple
+
+
+def run_doctor_checks() -> Tuple[bool, List[str]]:
+    """Runs system diagnostics for EvalRun environment.
+
+    Returns:
+        Tuple of (all_ok: bool, lines: List[str])
+    """
+    lines: List[str] = []
+    all_ok = True
+
+    lines.append("=====================================================================")
+    lines.append("                     EVALRUN SYSTEM DOCTOR                           ")
+    lines.append("=====================================================================")
+
+    # 1. Python Version Check
+    py_ver = sys.version.split()[0]
+    py_ok = sys.version_info >= (3, 10)
+    if not py_ok:
+        all_ok = False
+    lines.append(f"[{'PASS' if py_ok else 'FAIL'}] Python Version: {py_ver} (Requirement: >= 3.10)")
+
+    # 2. EvalRun Version & Installation State
+    try:
+        from importlib.metadata import PackageNotFoundError, version
+        pkg_version = version("evalrun")
+        install_type = "Installed Package"
+    except (PackageNotFoundError, ImportError):
+        pkg_version = "unknown"
+        pyproject = Path(__file__).resolve().parents[1] / "pyproject.toml"
+        try:
+            try:
+                import tomllib
+                with pyproject.open("rb") as f:
+                    metadata = tomllib.load(f)
+            except ModuleNotFoundError:
+                import re
+                metadata = {}
+                match = re.search(r'(?m)^version\s*=\s*["\']([^"\']+)["\']', pyproject.read_text(encoding="utf-8"))
+                if match:
+                    metadata = {"project": {"version": match.group(1)}}
+            pkg_version = str(metadata.get("project", {}).get("version", "unknown"))
+        except Exception:
+            pass
+        install_type = "Local Source Checkout"
+
+    lines.append(f"[PASS] EvalRun Version: {pkg_version} ({install_type})")
+
+    # 3. Model API Key Environment Variables
+    env_keys = {
+        "OPENAI_API_KEY": os.getenv("OPENAI_API_KEY"),
+        "GEMINI_API_KEY": os.getenv("GEMINI_API_KEY"),
+        "OPENROUTER_API_KEY": os.getenv("OPENROUTER_API_KEY"),
+        "NVIDIA_API_KEY": os.getenv("NVIDIA_API_KEY"),
+    }
+    set_keys = [k for k, v in env_keys.items() if v]
+    if set_keys:
+        lines.append(f"[PASS] Model API Keys Configured: {', '.join(set_keys)}")
+    else:
+        lines.append("[INFO] Model API Keys Configured: None (Offline demo works without keys)")
+
+    # 4. Agent Importability
+    try:
+        mod = importlib.import_module("agents.travel")
+        agent_cls = getattr(mod, "TravelPlanningAgent", None)
+        agent_ok = agent_cls is not None
+    except Exception:
+        agent_ok = False
+
+    if agent_ok:
+        lines.append("[PASS] Built-in Agent Import: agents.travel:TravelPlanningAgent")
+    else:
+        lines.append("[WARN] Built-in Agent Import: agents.travel not found in python path")
+
+    # 5. Endpoint Reachability (OpenAI API)
+    skip_network = os.getenv("EVALRUN_SKIP_NETWORK_CHECKS") == "1"
+    openai_reach = False
+    if not skip_network:
+        try:
+            req = urllib.request.Request("https://api.openai.com/v1/models", headers={"User-Agent": "evalrun-doctor"})
+            with urllib.request.urlopen(req, timeout=3) as resp:
+                openai_reach = resp.status in (200, 401)
+        except urllib.error.HTTPError as e:
+            openai_reach = e.code in (401, 403, 200)
+        except Exception:
+            openai_reach = False
+
+    if openai_reach:
+        lines.append("[PASS] Hosted Endpoint Reachability: https://api.openai.com/v1")
+    elif skip_network:
+        lines.append("[INFO] Hosted Endpoint Reachability: skipped (EVALRUN_SKIP_NETWORK_CHECKS=1)")
+    else:
+        lines.append("[INFO] Hosted Endpoint Reachability: https://api.openai.com/v1 (Offline or unreachable)")
+
+    # 6. Local Model Server Availability
+    local_reach = False
+    if not skip_network:
+        for local_url in ["http://localhost:8000/v1/models", "http://localhost:11434/api/tags"]:
+            try:
+                req = urllib.request.Request(local_url)
+                with urllib.request.urlopen(req, timeout=2) as resp:
+                    if resp.status in (200, 401, 403):
+                        local_reach = True
+                        break
+            except urllib.error.HTTPError as e:
+                if e.code in (200, 401, 403):
+                    local_reach = True
+                    break
+            except Exception:
+                pass
+
+    if local_reach:
+        lines.append("[PASS] Local Model Server: Active local LLM server detected")
+    elif skip_network:
+        lines.append("[INFO] Local Model Server: skipped (EVALRUN_SKIP_NETWORK_CHECKS=1)")
+    else:
+        lines.append("[INFO] Local Model Server: No active local LLM server detected on 8000/11434")
+
+    # 7. Write Access to Output Folder
+    output_dir = Path("./eval_results")
+    try:
+        output_dir.mkdir(parents=True, exist_ok=True)
+        test_file = output_dir / ".doctor_temp"
+        test_file.write_text("write_test", encoding="utf-8")
+        test_file.unlink()
+        write_ok = True
+    except Exception:
+        write_ok = False
+
+    if not write_ok:
+        all_ok = False
+    lines.append(f"[{'PASS' if write_ok else 'FAIL'}] Output Directory Write Access: {output_dir.resolve()}")
+
+    lines.append("=====================================================================")
+    if all_ok:
+        lines.append(" Doctor Verdict: SYSTEM READY FOR EVALUATION RUNS")
+    else:
+        lines.append(" Doctor Verdict: ISSUES DETECTED - PLEASE REVIEW WARNINGS ABOVE")
+    lines.append("=====================================================================")
+
+    return all_ok, lines