fix(testing/ci/tooling): consistent unittest, venv guidance, runnable lab commands

- #9: standardize the test chain on stdlib unittest (nothing-to-install, which
  keeps M13's claims true and its planted bug intact). Aligned M5/M14/M16 prose,
  M14 lab/test_tasks.py, and ci/gitlab starters; ruff stays the only pip install.
- #20: add venv / PEP 668 / which-python guidance to M20 (+ M14/M15 local
  installs); point MCP config at the venv's absolute python.
- #21: replace M21 Part D's empty `git diff HEAD~1` with `git log -p` (no
  .gitignore added — device preserved).
- #22: add a dependency-install step before M23's green baseline on a fresh clone.
- #23: M24 reviewer/triage now tolerate code-fence-wrapped JSON (stdlib only);
  feature.patch trap untouched.
- #28: fix M27 Part D CI snippet path (working-directory) and require the gate to
  target a varying candidate; swapped_model regression kept as the fixture.

Closes #9
Closes #20
Closes #21
Closes #22
Closes #23
Closes #28

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01TfzV5QvtPDz8LJS3Pu5VLT

modules/24-assistive-agents/lab/reviewer.py

@@ -19,6 +19,24 @@ from pathlib import Path
 
 HERE = Path(__file__).parent
 
+
+def load_json_response(path: Path):
+    """Parse the JSON the AI returned.
+
+    Chat assistants very often wrap their output in a ```json ... ``` code fence (or add a line of
+    prose) even when told to "return only the JSON" — so a strict json.loads on the raw paste fails
+    on the most likely real output. Try a strict parse first; if that fails, fall back to the
+    outermost { ... } block, which survives a code fence or surrounding text. Stdlib only."""
+    raw = path.read_text()
+    try:
+        return json.loads(raw)
+    except json.JSONDecodeError:
+        start, end = raw.find("{"), raw.rfind("}")
+        if start != -1 and end > start:
+            return json.loads(raw[start : end + 1])
+        raise
+
+
 PROMPT_HEADER = """\
 You are an assistive code reviewer. Follow the rubric below exactly, then review the diff that
 follows it. Return ONLY the JSON object the rubric specifies — no prose before or after.
@@ -40,7 +58,7 @@ def cmd_prompt(args: argparse.Namespace) -> int:
 
 def cmd_apply(args: argparse.Namespace) -> int:
     try:
-        review = json.loads(Path(args.response).read_text())
+        review = load_json_response(Path(args.response))
     except (json.JSONDecodeError, FileNotFoundError) as exc:
         print(f"error: could not read a JSON review from {args.response}: {exc}")
         return 1

modules/24-assistive-agents/lab/triage.py

@@ -39,6 +39,23 @@ def allowed_labels(taxonomy_text: str) -> set[str]:
     return set(LABEL_RE.findall(taxonomy_text))
 
 
+def load_json_response(path: Path):
+    """Parse the JSON the AI returned.
+
+    Chat assistants very often wrap their output in a ```json ... ``` code fence (or add a line of
+    prose) even when told to "return only the JSON" — so a strict json.loads on the raw paste fails
+    on the most likely real output. Try a strict parse first; if that fails, fall back to the
+    outermost { ... } block, which survives a code fence or surrounding text. Stdlib only."""
+    raw = path.read_text()
+    try:
+        return json.loads(raw)
+    except json.JSONDecodeError:
+        start, end = raw.find("{"), raw.rfind("}")
+        if start != -1 and end > start:
+            return json.loads(raw[start : end + 1])
+        raise
+
+
 def cmd_prompt(args: argparse.Namespace) -> int:
     taxonomy = Path(args.taxonomy).read_text()
     issue = Path(args.issue).read_text()
@@ -49,7 +66,7 @@ def cmd_prompt(args: argparse.Namespace) -> int:
 def cmd_apply(args: argparse.Namespace) -> int:
     allowed = allowed_labels(Path(args.taxonomy).read_text())
     try:
-        sug = json.loads(Path(args.response).read_text())
+        sug = load_json_response(Path(args.response))
     except (json.JSONDecodeError, FileNotFoundError) as exc:
         print(f"error: could not read a JSON suggestion from {args.response}: {exc}")
         return 1