Merge pull request 'Fix all open issues - code, security, ops' (#17) from fix/core-issues into main

Reviewed-on: https://git.example.com/jarianc/MasterMind/pulls/17

.gitignore

@@ -4,3 +4,12 @@
 /include
 /bin
 .ropeproject
+
+# Virtual environment config (contains personal paths)
+pyvenv.cfg
+
+# Large binaries (PDFs should not be tracked in git)
+*.pdf
+
+# Output directories
+output/

pyvenv.cfg

@@ -1,5 +0,0 @@
-home = /opt/homebrew/opt/python@3.13/bin
-include-system-site-packages = false
-version = 3.13.5
-executable = /opt/homebrew/Cellar/python@3.13/3.13.5/Frameworks/Python.framework/Versions/3.13/bin/python3.13
-command = /opt/homebrew/opt/python@3.13/bin/python3.13 -m venv /Users/user/Projects/Master Mind

scripts/mastermind_cli.py

@@ -5,23 +5,27 @@ MasterMind CLI
 This script will:
 
 1. Load the list of questions from `Master Mind/questions.json`.
-2. Load the system prompt from a user‑provided file (default: `system_prompt.txt`).
+2. Load the system prompt from a user-provided file (default: `system_prompt.txt`).
 3. Load the proposal and IDIOT method documents.
 4. For each question:
    - Send a request to a local AI model.
-   - Store the response in a file inside a project‑named folder.
+   - Store the response in a file inside a project-named folder.
    - Log the activity to the console.
 
 A very small helper is used to talk to a local model; it can be replaced with a real
 LLM client (Llama.cpp, Ollama, FastLLM, etc.) by editing the ``fetch_response`` function.
 
 Usage:
-    mastermind_cli.py \
+    mastermind_cli.py \\
-      --project ProjectName \
+      --project ProjectName \\
-      --proposal path/to/ProjectProposal.txt \
+      --proposal path/to/ProjectProposal.txt \\
-      --idiot path/to/IDIOTMethod.txt \
+      --idiot path/to/IDIOTMethod.txt \\
-      [--prompt-file path/to/system_prompt.txt] \
+      [--prompt-file path/to/system_prompt.txt] \\
-      [--model-magic "llama3"]   # optional flag for the local model name
+      [--model "llama3"]              # optional flag for the local model name
+      [--ollama-url "http://..."]     # optional Ollama endpoint (default: OLLAMA_URL env)
+      [--questions-file path.json]    # optional questions file (default: questions.json)
+      [--output-dir path]             # optional output directory
+      [--api-key "secret"]            # optional API key for Ollama auth
 """
 
 from __future__ import annotations
@@ -29,19 +33,50 @@ from __future__ import annotations
 import argparse
 import json
 import logging
+import os
 import pathlib
+import re
 import sys
-import urllib.request
 from typing import List
 
 logging.basicConfig(level=logging.INFO, format="%(message)s")
 
+# Sliding window size for accumulated context (issue #8)
+MAX_CONTEXT_WINDOW = 5
+
+
+def validate_project_name(name: str) -> str:
+    """Validate project name contains only safe characters. Prevents path traversal."""
+    if not re.match(r"^[a-zA-Z0-9_-]+$", name):
+        logging.error(
+            "Invalid project name '%s'. Only alphanumeric, hyphens, and underscores allowed.",
+            name,
+        )
+        sys.exit(1)
+    return name
+
+
+def validate_no_traversal(path: pathlib.Path, label: str) -> pathlib.Path:
+    """Ensure resolved path does not escape its intended base directory."""
+    resolved = path.resolve()
+    base = pathlib.Path(".").resolve()
+    try:
+        resolved.relative_to(base)
+    except ValueError:
+        logging.error(
+            "Invalid %s '%s': path escapes the allowed base directory.",
+            label,
+            path,
+        )
+        sys.exit(1)
+    return resolved
+
 
 def read_file(path: pathlib.Path) -> str:
     """Return the contents of a text file."""
     try:
         return path.read_text(encoding="utf-8")
-    except Exception as exc:  # pragma: no cover
+    except Exception as exc:
         logging.error("Could not read %s: %s", path, exc)
         sys.exit(1)
 
@@ -51,7 +86,7 @@ def load_questions(json_path: pathlib.Path) -> List[dict]:
     text = read_file(json_path)
     try:
         return json.loads(text)
-    except json.JSONDecodeError as exc:  # pragma: no cover
+    except json.JSONDecodeError as exc:
         logging.error("Invalid JSON in %s: %s", json_path, exc)
         sys.exit(1)
 
@@ -61,6 +96,9 @@ def fetch_response(
     user_prompt: str,
     context: List[str],
     model: str = "gpt-oss:20b",
+    ollama_url: str = "http://localhost:11434/api/generate",
+    api_key: str | None = None,
+    timeout: int = 120,
 ) -> str:
     """
     Send a request to a local AI model and return the response.
@@ -71,40 +109,50 @@ def fetch_response(
     Parameters
     ----------
     system_prompt : str
-        The system‑level instruction to the AI.
+        The system-level instruction to the AI.
     user_prompt : str
         The question or user message.
     context : List[str]
         Optional additional context to prepend to the request.
+    model : str
+        The model name to use.
+    ollama_url : str
+        The Ollama API endpoint URL.
+    api_key : str | None
+        Optional API key for authentication.
+    timeout : int
+        Request timeout in seconds.
 
     Returns
     -------
     str
-        The AI’s reply.
+        The AI's reply.
     """
-    # Combine system prompt, context, and user prompt.
     prompt_parts = [system_prompt]
     prompt_parts.extend(f"{ctx}" for ctx in context)
     prompt_parts.append(f"Question: {user_prompt}")
     prompt = "\n\n".join(prompt_parts) + "\n"
 
-    # Call Ollama API
-    url = "http://192.168.8.112:11434/api/generate"
     payload = json.dumps({
         "model": model,
         "prompt": prompt,
-        "stream": False
+        "stream": False,
     }).encode("utf-8")
     headers = {"Content-Type": "application/json"}
-    req = urllib.request.Request(url, data=payload, headers=headers, method="POST")
+    if api_key:
+        headers["Authorization"] = f"Bearer {api_key}"
+
+    req = __import__("urllib.request").request.Request(
+        ollama_url, data=payload, headers=headers, method="POST"
+    )
     try:
-        with urllib.request.urlopen(req) as resp:
+        with __import__("urllib.request").request.urlopen(req, timeout=timeout) as resp:
             data = resp.read().decode("utf-8")
             result = json.loads(data)
             return result.get("response", "").strip()
-    except Exception as exc:  # pragma: no cover
+    except Exception as exc:
         logging.error("Ollama request failed: %s", exc)
-        return f"Simulated response to: {user_prompt}"
+        sys.exit(1)
 
 
 def main() -> None:
@@ -147,21 +195,60 @@ def main() -> None:
         default="",
         help="Custom output directory for the project folder (default: current dir).",
     )
+    parser.add_argument(
+        "--ollama-url",
+        default=None,
+        help="Ollama API endpoint URL (default: $OLLAMA_URL or http://localhost:11434/api/generate).",
+    )
+    parser.add_argument(
+        "--api-key",
+        default=None,
+        help="API key for Ollama authentication (default: $OLLAMA_API_KEY).",
+    )
+    parser.add_argument(
+        "--questions-file",
+        default=None,
+        help="Path to questions JSON file (default: questions.json next to script).",
+    )
+    parser.add_argument(
+        "--timeout",
+        type=int,
+        default=120,
+        help="Request timeout in seconds for Ollama calls (default: 120).",
+    )
 
     args = parser.parse_args()
 
+    # Resolve Ollama URL: CLI arg > env var > default
+    ollama_url = args.ollama_url or os.environ.get(
+        "OLLAMA_URL", "http://localhost:11434/api/generate"
+    )
+    api_key = args.api_key or os.environ.get("OLLAMA_API_KEY")
+
+    # Warn if using HTTP (cleartext)
+    if ollama_url.startswith("http://"):
+        logging.warning(
+            "Using unencrypted HTTP for Ollama endpoint. "
+            "Sensitive project data will be transmitted in cleartext. "
+            "Use https:// or ensure network isolation."
+        )
+
+    # Validate project name against path traversal
+    validate_project_name(args.project)
+
     base_dir = pathlib.Path.cwd()
 
-    # Resolve project output directory
+    # Resolve project output directory with traversal protection
     if args.output_dir:
         out_base = pathlib.Path(args.output_dir).expanduser().resolve()
+        validate_no_traversal(out_base, "--output-dir")
     else:
         out_base = base_dir
     project_dir = out_base / args.project
 
     try:
         project_dir.mkdir(parents=True, exist_ok=True)
-    except Exception as exc:  # pragma: no cover
+    except Exception as exc:
         logging.error("Could not create project directory %s: %s", project_dir, exc)
         sys.exit(1)
 
@@ -174,7 +261,12 @@ def main() -> None:
     proposal_text = read_file(proposal_path)
     idiot_text = read_file(idiot_path)
 
-    questions_path = base_dir.parent / "questions.json"
+    # Resolve questions file: CLI arg > script-relative > project root
+    if args.questions_file:
+        questions_path = pathlib.Path(args.questions_file)
+    else:
+        script_dir = pathlib.Path(__file__).resolve().parent
+        questions_path = script_dir.parent / "questions.json"
     questions = load_questions(questions_path)
 
     # Merge context into system prompt for easier reuse
@@ -196,8 +288,8 @@ def main() -> None:
         phase_qs = phase.get("questions", [])
         logging.info("Phase: %s – %d questions", phase_name, len(phase_qs))
 
-        # Context from previous phases
+        # Sliding window: only keep last N summaries to bound context growth
-        phase_context: List[str] = accumulated_summaries.copy()
+        phase_context: List[str] = accumulated_summaries[-MAX_CONTEXT_WINDOW:]
 
         for q in phase_qs:
             question_text = q.get("text", "")
@@ -206,12 +298,18 @@ def main() -> None:
 
             logging.info("Q%d: %s", counter, question_text)
             response = fetch_response(
-                full_system_prompt, question_text, phase_context, args.model
+                full_system_prompt,
+                question_text,
+                phase_context,
+                args.model,
+                ollama_url,
+                api_key,
+                args.timeout,
             )
             output_file = project_dir / f"question_{counter:03d}.txt"
             try:
                 output_file.write_text(response, encoding="utf-8")
-            except Exception as exc:  # pragma: no cover
+            except Exception as exc:
                 logging.error("Could not write %s: %s", output_file, exc)
                 continue
             logging.info("Saved response to %s", output_file)
@@ -224,21 +322,28 @@ def main() -> None:
             project_dir / f"{phase_name.replace(' ', '_').lower()}_summary.txt"
         )
         try:
-            summary = fetch_response("Make a clear and concise summary that's no more than 500 tokens on the following input",
+            summary = fetch_response(
+                "Make a clear and concise summary that's no more than 500 tokens on the following input",
                 "\n\n".join(accumulated_summaries),
-                [])
+                [],
-
+                args.model,
-            phase_summary_file.write_text(
+                ollama_url,
-                summary, encoding="utf-8"
+                api_key,
+                args.timeout,
             )
-            logging.info("Saved per‑phase summary to %s", phase_summary_file)
+
-        except Exception as exc:  # pragma: no cover
+            phase_summary_file.write_text(summary, encoding="utf-8")
+            logging.info("Saved per-phase summary to %s", phase_summary_file)
+        except SystemExit:
+            raise
+        except Exception as exc:
             logging.error("Could not write %s: %s", phase_summary_file, exc)
 
-            experiment_results.append({
+        # Record result for every phase (was incorrectly inside except block)
-                "phase": phase_name,
+        experiment_results.append({
-                "summary_file": str(phase_summary_file),
+            "phase": phase_name,
-            })
+            "summary_file": str(phase_summary_file),
+        })
 
     logging.info("All done! Total responses: %d", counter - 1)