@jaguilar87/gaia-ops 3.10.2 → 3.11.0

package/hooks/pre_tool_use.py

@@ -152,6 +152,75 @@ def _load_agent_skills(subagent_type: str) -> str:
     return "\n\n---\n\n".join(parts) if parts else ""
 
 
+def _build_context_update_reminder(subagent_type: str) -> str:
+    """
+    Check which writable sections are empty and build a reminder.
+
+    Reads the context contracts to find writable sections for this agent,
+    then checks project-context.json to see which are empty.
+
+    Returns:
+        Reminder string or empty string if no empty sections.
+    """
+    if subagent_type not in PROJECT_AGENTS:
+        return ""
+
+    # Load contracts to find writable sections
+    contracts_paths = [
+        Path(".claude/config/context-contracts.gcp.json"),
+        Path(".claude/config/context-contracts.aws.json"),
+        Path(__file__).parent.parent / "config" / "context-contracts.gcp.json",
+        Path(__file__).parent.parent / "config" / "context-contracts.aws.json",
+    ]
+
+    writable = []
+    for cp in contracts_paths:
+        if cp.exists():
+            try:
+                data = json.loads(cp.read_text())
+                agent_perms = data.get("agents", {}).get(subagent_type, {})
+                writable = agent_perms.get("write", [])
+                if writable:
+                    break
+            except Exception:
+                continue
+
+    if not writable:
+        return ""
+
+    # Load project-context.json to find empty sections
+    pc_paths = [
+        Path(".claude/project-context/project-context.json"),
+        Path("project-context.json"),
+    ]
+
+    sections = {}
+    for pp in pc_paths:
+        if pp.exists():
+            try:
+                pc = json.loads(pp.read_text())
+                sections = pc.get("sections", {})
+                break
+            except Exception:
+                continue
+
+    # Find empty writable sections
+    empty = []
+    for section_name in writable:
+        section_data = sections.get(section_name, {})
+        if not section_data or section_data == {}:
+            empty.append(section_name)
+
+    if not empty:
+        return ""
+
+    empty_list = ", ".join(f"`{s}`" for s in empty)
+    return (
+        f"\n**CONTEXT_UPDATE REQUIRED:** Your writable sections {empty_list} "
+        f"are currently EMPTY. After completing your task, you MUST emit a "
+        f"CONTEXT_UPDATE block with any data you discovered. "
+        f"See \"Context Updater Protocol\" above for the format.\n\n"
+    )
 
 
 def _should_inject_on_resume(parameters: dict) -> bool:
@@ -339,13 +408,16 @@ def _inject_project_context(parameters: dict) -> dict:
         skills_content = _load_agent_skills(subagent_type)
         skills_section = f"\n\n---\n\n# Agent Skills (Auto-Injected)\n\n{skills_content}" if skills_content else ""
 
+        # Build context update reminder for empty writable sections
+        update_reminder = _build_context_update_reminder(subagent_type)
+
         # Inject context and skills into prompt
         enriched_prompt = f"""# Project Context (Auto-Injected)
 
 {json.dumps(context_payload, indent=2)}
 
 {pending_warning}---{skills_section}
-
+{update_reminder}
 # User Task
 
 {prompt}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@jaguilar87/gaia-ops",
-  "version": "3.10.2",
+  "version": "3.11.0",
   "description": "Multi-agent orchestration system for Claude Code - DevOps automation toolkit",
   "main": "index.js",
   "type": "module",
@@ -54,7 +54,12 @@
     "index.js"
   ],
   "scripts": {
-    "test": "python3 -m pytest tests/ -v",
+    "test": "python3 -m pytest tests/ -v --ignore=tests/layer2_llm_evaluation --ignore=tests/layer3_e2e",
+    "test:layer1": "python3 -m pytest tests/ -v --ignore=tests/layer2_llm_evaluation --ignore=tests/layer3_e2e",
+    "test:layer2": "python3 -m pytest tests/layer2_llm_evaluation/ -v -m llm",
+    "test:layer3": "python3 -m pytest tests/layer3_e2e/ -v -m e2e",
+    "test:all": "python3 -m pytest tests/ -v -m ''",
+    "test:promptfoo": "npx promptfoo eval --config tests/promptfoo.yaml",
     "lint": "eslint .",
     "clean": "find . -type d -name '__pycache__' -exec rm -rf {} + 2>/dev/null || true",
     "pre-publish": "node bin/pre-publish-validate.js",

package/tests/conftest.py

@@ -0,0 +1,166 @@
+"""
+Root conftest.py - Shared test infrastructure for gaia-ops.
+
+Provides:
+- Custom markers: llm, e2e (auto-skipped in default test runs)
+- Session fixtures: package_root, agents_dir, skills_dir, config_dir, hooks_dir
+- Frontmatter parser (manual, no PyYAML dependency)
+"""
+
+import pytest
+from pathlib import Path
+
+
+# ============================================================================
+# MARKERS
+# ============================================================================
+
+def pytest_configure(config):
+    """Register custom markers."""
+    config.addinivalue_line("markers", "llm: LLM evaluation tests (require ANTHROPIC_API_KEY)")
+    config.addinivalue_line("markers", "e2e: E2E headless tests (require claude CLI)")
+
+
+def pytest_collection_modifyitems(config, items):
+    """Auto-skip llm and e2e tests unless explicitly requested via -m flag."""
+    # If user explicitly passed -m, respect that
+    markexpr = config.getoption("-m", default="")
+    if markexpr:
+        return
+
+    skip_llm = pytest.mark.skip(reason="LLM tests skipped by default (use -m llm)")
+    skip_e2e = pytest.mark.skip(reason="E2E tests skipped by default (use -m e2e)")
+
+    for item in items:
+        if "llm" in item.keywords:
+            item.add_marker(skip_llm)
+        if "e2e" in item.keywords:
+            item.add_marker(skip_e2e)
+
+
+# ============================================================================
+# SESSION FIXTURES
+# ============================================================================
+
+@pytest.fixture(scope="session")
+def package_root():
+    """Root of the gaia-ops package."""
+    root = Path(__file__).resolve().parents[1]
+    return root.resolve() if root.is_symlink() else root
+
+
+@pytest.fixture(scope="session")
+def agents_dir(package_root):
+    """Directory containing agent definition .md files."""
+    d = package_root / "agents"
+    return d.resolve() if d.is_symlink() else d
+
+
+@pytest.fixture(scope="session")
+def skills_dir(package_root):
+    """Directory containing skill directories with SKILL.md files."""
+    d = package_root / "skills"
+    return d.resolve() if d.is_symlink() else d
+
+
+@pytest.fixture(scope="session")
+def config_dir(package_root):
+    """Directory containing config files (context-contracts, etc)."""
+    d = package_root / "config"
+    return d.resolve() if d.is_symlink() else d
+
+
+@pytest.fixture(scope="session")
+def hooks_dir(package_root):
+    """Directory containing hook scripts."""
+    d = package_root / "hooks"
+    return d.resolve() if d.is_symlink() else d
+
+
+@pytest.fixture(scope="session")
+def claude_md_content(package_root):
+    """Content of the orchestrator CLAUDE.md."""
+    return (package_root / "CLAUDE.md").read_text()
+
+
+@pytest.fixture(scope="session")
+def all_agent_files(agents_dir):
+    """All agent .md files (excluding READMEs)."""
+    return [f for f in agents_dir.glob("*.md") if "README" not in f.name.upper()]
+
+
+@pytest.fixture(scope="session")
+def all_skill_dirs(skills_dir):
+    """All skill directories that contain a SKILL.md."""
+    return [d for d in skills_dir.iterdir() if d.is_dir() and (d / "SKILL.md").exists()]
+
+
+# ============================================================================
+# FRONTMATTER PARSER (manual, no PyYAML)
+# ============================================================================
+
+def parse_frontmatter(text):
+    """
+    Parse YAML frontmatter from markdown text (manual parser, no PyYAML).
+
+    Supports simple key-value pairs and lists (- item).
+
+    Args:
+        text: Full markdown text starting with ---
+
+    Returns:
+        dict with parsed frontmatter fields, or empty dict if no frontmatter
+    """
+    if not text.startswith("---"):
+        return {}
+
+    try:
+        end = text.index("---", 3)
+    except ValueError:
+        return {}
+
+    fm_text = text[3:end]
+    result = {}
+    current_key = None
+    current_list = None
+
+    for line in fm_text.splitlines():
+        stripped = line.strip()
+        if not stripped or stripped.startswith("#"):
+            continue
+
+        # List item under current key
+        if stripped.startswith("- ") and current_key and current_list is not None:
+            current_list.append(stripped[2:].strip())
+            continue
+
+        # New key-value pair
+        if ":" in stripped:
+            # End previous list
+            if current_key and current_list is not None:
+                result[current_key] = current_list
+
+            key, _, value = stripped.partition(":")
+            key = key.strip()
+            value = value.strip()
+
+            if value:
+                result[key] = value
+                current_key = key
+                current_list = None
+            else:
+                # Start of a list
+                current_key = key
+                current_list = []
+        else:
+            # Not a key-value, not a list item - end list
+            if current_key and current_list is not None:
+                result[current_key] = current_list
+                current_key = None
+                current_list = None
+
+    # Finalize last list
+    if current_key and current_list is not None:
+        result[current_key] = current_list
+
+    return result