Add support for AGENTS.md files in microagent system (#10528)

Co-authored-by: openhands <openhands@all-hands.dev>

openhands/microagent/microagent.py

@@ -2,7 +2,7 @@ import io
 import re
 from itertools import chain
 from pathlib import Path
-from typing import Union
+from typing import ClassVar, Union
 
 import frontmatter
 from pydantic import BaseModel
@@ -23,6 +23,31 @@ class BaseMicroagent(BaseModel):
     source: str  # path to the file
     type: MicroagentType
 
+    PATH_TO_THIRD_PARTY_MICROAGENT_NAME: ClassVar[dict[str, str]] = {
+        '.cursorrules': 'cursorrules',
+        'agents.md': 'agents',
+        'agent.md': 'agents',
+    }
+
+    @classmethod
+    def _handle_third_party(
+        cls, path: Path, file_content: str
+    ) -> Union['RepoMicroagent', None]:
+        # Determine the agent name based on file type
+        microagent_name = cls.PATH_TO_THIRD_PARTY_MICROAGENT_NAME.get(path.name.lower())
+
+        # Create RepoMicroagent if we recognized the file type
+        if microagent_name is not None:
+            return RepoMicroagent(
+                name=microagent_name,
+                content=file_content,
+                metadata=MicroagentMetadata(name=microagent_name),
+                source=str(path),
+                type=MicroagentType.REPO_KNOWLEDGE,
+            )
+
+        return None
+
     @classmethod
     def load(
         cls,
@@ -40,11 +65,10 @@ class BaseMicroagent(BaseModel):
         # Otherwise, we will rely on the name from metadata later
         derived_name = None
         if microagent_dir is not None:
-            # Special handling for .cursorrules files which are not in microagent_dir
-            if path.name == '.cursorrules':
-                derived_name = 'cursorrules'
-            else:
-                derived_name = str(path.relative_to(microagent_dir).with_suffix(''))
+            # Special handling for files which are not in microagent_dir
+            derived_name = cls.PATH_TO_THIRD_PARTY_MICROAGENT_NAME.get(
+                path.name.lower()
+            ) or str(path.relative_to(microagent_dir).with_suffix(''))
 
         # Only load directly from path if file_content is not provided
         if file_content is None:
@@ -61,15 +85,10 @@ class BaseMicroagent(BaseModel):
                 type=MicroagentType.REPO_KNOWLEDGE,
             )
 
-        # Handle .cursorrules files
-        if path.name == '.cursorrules':
-            return RepoMicroagent(
-                name='cursorrules',
-                content=file_content,
-                metadata=MicroagentMetadata(name='cursorrules'),
-                source=str(path),
-                type=MicroagentType.REPO_KNOWLEDGE,
-            )
+        # Handle third-party agent instruction files
+        third_party_agent = cls._handle_third_party(path, file_content)
+        if third_party_agent is not None:
+            return third_party_agent
 
         file_io = io.StringIO(file_content)
         loaded = frontmatter.load(file_io)
@@ -276,31 +295,44 @@ def load_microagents_from_dir(
 
     # Load all agents from microagents directory
     logger.debug(f'Loading agents from {microagent_dir}')
-    if microagent_dir.exists():
-        # Collect .cursorrules file from repo root and .md files from microagents dir
-        cursorrules_files = []
-        if (microagent_dir.parent.parent / '.cursorrules').exists():
-            cursorrules_files = [microagent_dir.parent.parent / '.cursorrules']
 
+    # Always check for .cursorrules and AGENTS.md files in repo root, regardless of whether microagents_dir exists
+    special_files = []
+    repo_root = microagent_dir.parent.parent
+
+    # Check for .cursorrules
+    if (repo_root / '.cursorrules').exists():
+        special_files.append(repo_root / '.cursorrules')
+
+    # Check for AGENTS.md (case-insensitive)
+    for agents_filename in ['AGENTS.md', 'agents.md', 'AGENT.md', 'agent.md']:
+        agents_path = repo_root / agents_filename
+        if agents_path.exists():
+            special_files.append(agents_path)
+            break  # Only add the first one found to avoid duplicates
+
+    # Collect .md files from microagents directory if it exists
+    md_files = []
+    if microagent_dir.exists():
         md_files = [f for f in microagent_dir.rglob('*.md') if f.name != 'README.md']
 
-        # Process all files in one loop
-        for file in chain(cursorrules_files, md_files):
-            try:
-                agent = BaseMicroagent.load(file, microagent_dir)
-                if isinstance(agent, RepoMicroagent):
-                    repo_agents[agent.name] = agent
-                elif isinstance(agent, KnowledgeMicroagent):
-                    # Both KnowledgeMicroagent and TaskMicroagent go into knowledge_agents
-                    knowledge_agents[agent.name] = agent
-            except MicroagentValidationError as e:
-                # For validation errors, include the original exception
-                error_msg = f'Error loading microagent from {file}: {str(e)}'
-                raise MicroagentValidationError(error_msg) from e
-            except Exception as e:
-                # For other errors, wrap in a ValueError with detailed message
-                error_msg = f'Error loading microagent from {file}: {str(e)}'
-                raise ValueError(error_msg) from e
+    # Process all files in one loop
+    for file in chain(special_files, md_files):
+        try:
+            agent = BaseMicroagent.load(file, microagent_dir)
+            if isinstance(agent, RepoMicroagent):
+                repo_agents[agent.name] = agent
+            elif isinstance(agent, KnowledgeMicroagent):
+                # Both KnowledgeMicroagent and TaskMicroagent go into knowledge_agents
+                knowledge_agents[agent.name] = agent
+        except MicroagentValidationError as e:
+            # For validation errors, include the original exception
+            error_msg = f'Error loading microagent from {file}: {str(e)}'
+            raise MicroagentValidationError(error_msg) from e
+        except Exception as e:
+            # For other errors, wrap in a ValueError with detailed message
+            error_msg = f'Error loading microagent from {file}: {str(e)}'
+            raise ValueError(error_msg) from e
 
     logger.debug(
         f'Loaded {len(repo_agents) + len(knowledge_agents)} microagents: '

tests/unit/microagent/test_microagent_utils.py

@@ -364,3 +364,184 @@ def test_load_microagents_with_cursorrules(temp_microagents_dir_with_cursorrules
     assert cursorrules_agent.name == 'cursorrules'
     assert 'Always use TypeScript for new files' in cursorrules_agent.content
     assert cursorrules_agent.type == MicroagentType.REPO_KNOWLEDGE
+
+
+@pytest.fixture
+def temp_dir_with_cursorrules_only():
+    """Create a temporary directory with only .cursorrules file (no .openhands/microagents directory)."""
+    with tempfile.TemporaryDirectory() as temp_dir:
+        root = Path(temp_dir)
+
+        # Create .cursorrules file in repository root
+        cursorrules_content = """Always use Python for new files.
+Follow PEP 8 style guidelines."""
+        (root / '.cursorrules').write_text(cursorrules_content)
+
+        # Note: We intentionally do NOT create .openhands/microagents directory
+        yield root
+
+
+def test_load_cursorrules_without_microagents_dir(temp_dir_with_cursorrules_only):
+    """Test loading .cursorrules file when .openhands/microagents directory doesn't exist.
+
+    This test reproduces the bug where .cursorrules is only loaded when
+    .openhands/microagents directory exists.
+    """
+    # Try to load from non-existent microagents directory
+    microagents_dir = temp_dir_with_cursorrules_only / '.openhands' / 'microagents'
+
+    repo_agents, knowledge_agents = load_microagents_from_dir(microagents_dir)
+
+    # This should find the .cursorrules file even though microagents_dir doesn't exist
+    assert len(repo_agents) == 1  # Only .cursorrules
+    assert 'cursorrules' in repo_agents
+    assert len(knowledge_agents) == 0
+
+    # Check .cursorrules agent
+    cursorrules_agent = repo_agents['cursorrules']
+    assert isinstance(cursorrules_agent, RepoMicroagent)
+    assert cursorrules_agent.name == 'cursorrules'
+    assert 'Always use Python for new files' in cursorrules_agent.content
+    assert cursorrules_agent.type == MicroagentType.REPO_KNOWLEDGE
+
+
+def test_agents_md_file_load():
+    """Test loading AGENTS.md file as a RepoMicroagent."""
+    agents_content = """# Project Setup
+
+## Setup commands
+
+- Install deps: `npm install`
+- Start dev server: `npm run dev`
+- Run tests: `npm test`
+
+## Code style
+
+- TypeScript strict mode
+- Single quotes, no semicolons
+- Use functional patterns where possible"""
+
+    agents_path = Path('AGENTS.md')
+
+    # Test loading AGENTS.md file directly
+    agent = BaseMicroagent.load(agents_path, file_content=agents_content)
+
+    # Verify it's loaded as a RepoMicroagent
+    assert isinstance(agent, RepoMicroagent)
+    assert agent.name == 'agents'
+    assert agent.content == agents_content
+    assert agent.type == MicroagentType.REPO_KNOWLEDGE
+    assert agent.metadata.name == 'agents'
+    assert agent.source == str(agents_path)
+
+
+def test_agents_md_case_insensitive():
+    """Test that AGENTS.md loading is case-insensitive."""
+    agents_content = """# Development Guide
+
+Use TypeScript for all new files."""
+
+    test_cases = ['AGENTS.md', 'agents.md', 'AGENT.md', 'agent.md']
+
+    for filename in test_cases:
+        agents_path = Path(filename)
+        agent = BaseMicroagent.load(agents_path, file_content=agents_content)
+
+        assert isinstance(agent, RepoMicroagent)
+        assert agent.name == 'agents'
+        assert agent.content == agents_content
+        assert agent.type == MicroagentType.REPO_KNOWLEDGE
+
+
+@pytest.fixture
+def temp_dir_with_agents_md_only():
+    """Create a temporary directory with only AGENTS.md file (no .openhands/microagents directory)."""
+    with tempfile.TemporaryDirectory() as temp_dir:
+        root = Path(temp_dir)
+
+        # Create AGENTS.md file in repository root
+        agents_content = """# Development Guide
+
+## Setup commands
+
+- Install deps: `poetry install`
+- Start dev server: `poetry run python app.py`
+- Run tests: `poetry run pytest`
+
+## Code style
+
+- Python 3.12+
+- Follow PEP 8 guidelines
+- Use type hints everywhere"""
+        (root / 'AGENTS.md').write_text(agents_content)
+
+        # Note: We intentionally do NOT create .openhands/microagents directory
+        yield root
+
+
+def test_load_agents_md_without_microagents_dir(temp_dir_with_agents_md_only):
+    """Test loading AGENTS.md file when .openhands/microagents directory doesn't exist."""
+    # Try to load from non-existent microagents directory
+    microagents_dir = temp_dir_with_agents_md_only / '.openhands' / 'microagents'
+
+    repo_agents, knowledge_agents = load_microagents_from_dir(microagents_dir)
+
+    # This should find the AGENTS.md file even though microagents_dir doesn't exist
+    assert len(repo_agents) == 1  # Only AGENTS.md
+    assert 'agents' in repo_agents
+    assert len(knowledge_agents) == 0
+
+    # Check AGENTS.md agent
+    agents_agent = repo_agents['agents']
+    assert isinstance(agents_agent, RepoMicroagent)
+    assert agents_agent.name == 'agents'
+    assert 'Install deps: `poetry install`' in agents_agent.content
+    assert agents_agent.type == MicroagentType.REPO_KNOWLEDGE
+
+
+@pytest.fixture
+def temp_dir_with_both_cursorrules_and_agents():
+    """Create a temporary directory with both .cursorrules and AGENTS.md files."""
+    with tempfile.TemporaryDirectory() as temp_dir:
+        root = Path(temp_dir)
+
+        # Create .cursorrules file
+        cursorrules_content = """Always use Python for new files.
+Follow PEP 8 style guidelines."""
+        (root / '.cursorrules').write_text(cursorrules_content)
+
+        # Create AGENTS.md file
+        agents_content = """# Development Guide
+
+## Setup commands
+
+- Install deps: `poetry install`
+- Run tests: `poetry run pytest`"""
+        (root / 'AGENTS.md').write_text(agents_content)
+
+        yield root
+
+
+def test_load_both_cursorrules_and_agents_md(temp_dir_with_both_cursorrules_and_agents):
+    """Test loading both .cursorrules and AGENTS.md files when .openhands/microagents doesn't exist."""
+    # Try to load from non-existent microagents directory
+    microagents_dir = (
+        temp_dir_with_both_cursorrules_and_agents / '.openhands' / 'microagents'
+    )
+
+    repo_agents, knowledge_agents = load_microagents_from_dir(microagents_dir)
+
+    # This should find both files
+    assert len(repo_agents) == 2  # .cursorrules + AGENTS.md
+    assert 'cursorrules' in repo_agents
+    assert 'agents' in repo_agents
+    assert len(knowledge_agents) == 0
+
+    # Check both agents
+    cursorrules_agent = repo_agents['cursorrules']
+    assert isinstance(cursorrules_agent, RepoMicroagent)
+    assert 'Always use Python for new files' in cursorrules_agent.content
+
+    agents_agent = repo_agents['agents']
+    assert isinstance(agents_agent, RepoMicroagent)
+    assert 'Install deps: `poetry install`' in agents_agent.content