drydock-cli 0.6.5__tar.gz → 0.6.7__tar.gz

{drydock_cli-0.6.5 → drydock_cli-0.6.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: drydock-cli
-Version: 0.6.5
+Version: 0.6.7
 Summary: Drydock — nautical CLI coding agent
 Project-URL: Homepage, https://github.com/fbobe321/drydock
 Project-URL: Repository, https://github.com/fbobe321/drydock

{drydock_cli-0.6.5 → drydock_cli-0.6.7}/drydock/core/agent_loop.py

@@ -849,8 +849,9 @@ class AgentLoop:
 
     def _circuit_breaker_check(self, tool_call: ResolvedToolCall) -> str | None:
         """Block exact-duplicate tool calls. Returns cached result or None."""
+        args_str = json.dumps(tool_call.args_dict, sort_keys=True, default=str)
         sig = hashlib.md5(
-            f"{tool_call.tool_name}:{tool_call.raw_arguments}".encode()
+            f"{tool_call.tool_name}:{args_str}".encode()
         ).hexdigest()
 
         count, last_result = self._tool_call_history.get(sig, (0, ""))
@@ -883,8 +884,9 @@ class AgentLoop:
 
     def _circuit_breaker_record(self, tool_call: ResolvedToolCall, result_text: str) -> None:
         """Record a tool call execution for circuit breaker tracking."""
+        args_str = json.dumps(tool_call.args_dict, sort_keys=True, default=str)
         sig = hashlib.md5(
-            f"{tool_call.tool_name}:{tool_call.raw_arguments}".encode()
+            f"{tool_call.tool_name}:{args_str}".encode()
         ).hexdigest()
         count, _ = self._tool_call_history.get(sig, (0, ""))
         self._tool_call_history[sig] = (count + 1, result_text[:500])

{drydock_cli-0.6.5 → drydock_cli-0.6.7}/drydock/core/tools/builtins/webfetch.py

@@ -23,7 +23,7 @@ if TYPE_CHECKING:
     from drydock.core.types import ToolCallEvent, ToolResultEvent
 
 
-_HONEST_USER_AGENT = "vibe-cli"
+_HONEST_USER_AGENT = "drydock-cli"
 _HTTP_FORBIDDEN = 403
 
 
@@ -146,8 +146,10 @@ class WebFetch(
     async def _do_fetch(
         self, url: str, timeout: int, headers: dict[str, str]
     ) -> httpx.Response:
+        import os
+        verify_ssl = os.environ.get("DRYDOCK_INSECURE") != "1"
         async with httpx.AsyncClient(
-            follow_redirects=True, timeout=httpx.Timeout(timeout)
+            follow_redirects=True, timeout=httpx.Timeout(timeout), verify=verify_ssl
         ) as client:
             response = await client.get(url, headers=headers)
 

{drydock_cli-0.6.5 → drydock_cli-0.6.7}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "drydock-cli"
-version = "0.6.5"
+version = "0.6.7"
 description = "Drydock — nautical CLI coding agent"
 readme = "README.md"
 requires-python = ">=3.12"

{drydock_cli-0.6.5 → drydock_cli-0.6.7}/scripts/deploy_to_github.sh

@@ -37,7 +37,7 @@ AUTH_URL="https://${GITHUB_TOKEN}@github.com/fbobe321/drydock.git"
 
 log "Running regression tests..."
 cd "$DRYDOCK_SRC"
-if ! python3 -m pytest tests/test_drydock_regression.py tests/test_drydock_tasks.py tests/test_loop_detection.py \
+if ! python3 -m pytest tests/test_drydock_regression.py tests/test_drydock_tasks.py tests/test_loop_detection.py tests/test_agent_tasks.py \
     -p no:xdist -p no:cov --override-ini="addopts=" -q 2>&1; then
     log "TESTS FAILED — deploy aborted. Fix the tests before deploying."
     exit 1

{drydock_cli-0.6.5 → drydock_cli-0.6.7}/scripts/publish_to_pypi.sh

@@ -25,7 +25,7 @@ PYPI_TOKEN=$(cat "$TOKEN_FILE")
 
 # --- Run tests ---
 log "Running regression tests..."
-if ! $PYTHON -m pytest tests/test_drydock_regression.py tests/test_drydock_tasks.py tests/test_loop_detection.py \
+if ! $PYTHON -m pytest tests/test_drydock_regression.py tests/test_drydock_tasks.py tests/test_loop_detection.py tests/test_agent_tasks.py \
     -p no:xdist -p no:cov --override-ini="addopts=" -q 2>&1; then
     log "TESTS FAILED — publish aborted."
     exit 1

drydock_cli-0.6.7/tests/test_agent_tasks.py

@@ -0,0 +1,238 @@
+"""Agent task tests — spin up real agent loops with scripted LLM responses.
+
+Each test simulates a real user task by providing scripted model responses
+and verifying the agent produces the right events and behavior.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import json
+
+import pytest
+
+from tests.conftest import build_test_agent_loop, build_test_vibe_config
+from tests.mock.utils import mock_llm_chunk
+from tests.stubs.fake_backend import FakeBackend
+from drydock.core.agent_loop import AgentLoop
+from drydock.core.agents.models import BuiltinAgentName
+from drydock.core.tools.base import BaseToolConfig, ToolPermission
+from drydock.core.types import (
+    AssistantEvent,
+    BaseEvent,
+    FunctionCall,
+    LLMMessage,
+    MessageList,
+    Role,
+    ToolCall,
+    ToolCallEvent,
+    ToolResultEvent,
+    UserMessageEvent,
+)
+
+
+async def _run(agent: AgentLoop, prompt: str, max_events: int = 50) -> list[BaseEvent]:
+    events = []
+    async for ev in agent.act(prompt):
+        events.append(ev)
+        if len(events) >= max_events:
+            break
+    return events
+
+
+def _count(events: list[BaseEvent], event_type: type) -> int:
+    return sum(1 for e in events if isinstance(e, event_type))
+
+
+def _agent(backend: FakeBackend) -> AgentLoop:
+    config = build_test_vibe_config(
+        system_prompt_id="tests",
+        include_project_context=False,
+        include_prompt_detail=False,
+    )
+    return build_test_agent_loop(
+        config=config,
+        agent_name=BuiltinAgentName.AUTO_APPROVE,
+        backend=backend,
+    )
+
+
+# ============================================================================
+# Task 1: Simple greeting — no tools, no loops
+# ============================================================================
+
+@pytest.mark.asyncio
+async def test_task_greeting_no_loop(telemetry_events):
+    """Agent responds to greeting and stops without looping."""
+    backend = FakeBackend([
+        [mock_llm_chunk(content="Hello! How can I help?")],
+    ])
+    events = await _run(_agent(backend), "hi")
+
+    assert _count(events, UserMessageEvent) == 1
+    text = [e for e in events if isinstance(e, AssistantEvent)]
+    assert len(text) >= 1
+    assert "Hello" in text[0].content
+    assert len(events) < 10  # Not looping
+
+
+# ============================================================================
+# Task 2: Agent stops after text-only response
+# ============================================================================
+
+@pytest.mark.asyncio
+async def test_task_stops_on_text_response(telemetry_events):
+    """Text-only response terminates the loop (no tool calls = done)."""
+    backend = FakeBackend([
+        [mock_llm_chunk(content="The answer is 42.")],
+    ])
+    events = await _run(_agent(backend), "meaning of life?")
+
+    assert _count(events, AssistantEvent) >= 1
+    assert len(events) <= 5
+
+
+# ============================================================================
+# Task 3: Empty response doesn't crash
+# ============================================================================
+
+@pytest.mark.asyncio
+async def test_task_empty_response_no_crash(telemetry_events):
+    """Empty LLM response handled gracefully."""
+    backend = FakeBackend([[mock_llm_chunk(content="")]])
+    events = await _run(_agent(backend), "hello")
+    assert _count(events, UserMessageEvent) == 1
+
+
+# ============================================================================
+# Task 4: Message ordering maintained across turns
+# ============================================================================
+
+@pytest.mark.asyncio
+async def test_task_message_ordering_clean(telemetry_events):
+    """No user-after-tool violations in message history."""
+    backend = FakeBackend([
+        [mock_llm_chunk(content="First response.")],
+    ])
+    agent = _agent(backend)
+    await _run(agent, "test message ordering")
+
+    for i in range(1, len(agent.messages)):
+        assert not (
+            agent.messages[i].role == Role.user
+            and agent.messages[i - 1].role == Role.tool
+        ), f"user after tool at position {i}"
+
+
+# ============================================================================
+# Task 5: Messages remain a MessageList (not plain list)
+# ============================================================================
+
+@pytest.mark.asyncio
+async def test_task_messagelist_integrity(telemetry_events):
+    """Agent messages stay as MessageList, never replaced with plain list."""
+    backend = FakeBackend([
+        [mock_llm_chunk(content="Working on it.")],
+    ])
+    agent = _agent(backend)
+    await _run(agent, "test integrity")
+
+    assert isinstance(agent.messages, MessageList)
+    assert hasattr(agent.messages, "reset")
+    assert hasattr(agent.messages, "append")
+
+
+# ============================================================================
+# Task 6: Circuit breaker state initialized
+# ============================================================================
+
+@pytest.mark.asyncio
+async def test_task_circuit_breaker_initialized(telemetry_events):
+    """Agent has circuit breaker history dict after creation."""
+    backend = FakeBackend([[mock_llm_chunk(content="ok")]])
+    agent = _agent(backend)
+    assert hasattr(agent, "_tool_call_history")
+    assert isinstance(agent._tool_call_history, dict)
+
+
+# ============================================================================
+# Task 7: Middleware turn limit stops agent
+# ============================================================================
+
+@pytest.mark.asyncio
+async def test_task_turn_limit_stops_agent(telemetry_events):
+    """max_turns=1 limits agent to a single turn."""
+    backend = FakeBackend([
+        [mock_llm_chunk(content="First turn.")],
+        [mock_llm_chunk(content="Should not reach this.")],
+    ])
+    config = build_test_vibe_config(
+        system_prompt_id="tests",
+        include_project_context=False,
+        include_prompt_detail=False,
+    )
+    agent = build_test_agent_loop(
+        config=config,
+        agent_name=BuiltinAgentName.AUTO_APPROVE,
+        backend=backend,
+        max_turns=1,
+    )
+    events = await _run(agent, "do many things")
+
+    # With max_turns=1, agent should stop after 1 turn
+    # Either via middleware stop or by natural text-only response
+    text = [e for e in events if isinstance(e, AssistantEvent)]
+    assert len(text) >= 1
+    assert len(events) <= 10
+
+
+# ============================================================================
+# Task 8: Multi-response conversation
+# ============================================================================
+
+@pytest.mark.asyncio
+async def test_task_multi_turn_conversation(telemetry_events):
+    """Agent handles multiple text-only turns (when nudged to continue)."""
+    backend = FakeBackend([
+        [mock_llm_chunk(content="Let me think about this...")],
+    ])
+    agent = _agent(backend)
+    events = await _run(agent, "explain the codebase")
+
+    # Should have at least user message + 1 assistant response
+    assert _count(events, UserMessageEvent) == 1
+    assert _count(events, AssistantEvent) >= 1
+
+
+# ============================================================================
+# Task 9: Backend exception doesn't crash permanently
+# ============================================================================
+
+@pytest.mark.asyncio
+async def test_task_backend_error_recovery(telemetry_events):
+    """Agent survives an LLM backend error."""
+    backend = FakeBackend(
+        exception_to_raise=RuntimeError("API error from vllm: connection refused")
+    )
+    agent = _agent(backend)
+
+    # Should not raise — agent catches and handles
+    events = await _run(agent, "test error handling")
+    # Will have limited events since backend keeps failing
+    assert _count(events, UserMessageEvent) == 1
+
+
+# ============================================================================
+# Task 10: System prompt present in messages
+# ============================================================================
+
+@pytest.mark.asyncio
+async def test_task_system_prompt_present(telemetry_events):
+    """Agent has a system prompt as first message."""
+    backend = FakeBackend([[mock_llm_chunk(content="ok")]])
+    agent = _agent(backend)
+    await _run(agent, "test")
+
+    assert len(agent.messages) >= 2
+    assert agent.messages[0].role == Role.system
+    assert len(agent.messages[0].content) > 50  # Non-trivial system prompt

{drydock_cli-0.6.5 → drydock_cli-0.6.7}/tests/test_drydock_tasks.py

@@ -400,7 +400,7 @@ class TestCircuitBreaker:
         al.messages = MessageList()
         al._tool_call_history = {}
 
-        tc = SimpleNamespace(tool_name="bash", raw_arguments='{"command": "ls -ltr"}')
+        tc = SimpleNamespace(tool_name="bash", args_dict={"command": "ls -ltr"})
 
         assert al._circuit_breaker_check(tc) is None
         al._circuit_breaker_record(tc, "file1.py\nfile2.py")
@@ -421,8 +421,8 @@ class TestCircuitBreaker:
         al.messages = MessageList()
         al._tool_call_history = {}
 
-        tc1 = SimpleNamespace(tool_name="bash", raw_arguments='{"command": "ls -ltr"}')
-        tc2 = SimpleNamespace(tool_name="bash", raw_arguments='{"command": "ls -la"}')
+        tc1 = SimpleNamespace(tool_name="bash", args_dict={"command": "ls -ltr"})
+        tc2 = SimpleNamespace(tool_name="bash", args_dict={"command": "ls -la"})
 
         al._circuit_breaker_record(tc1, "result1")
         al._circuit_breaker_record(tc1, "result1")

{drydock_cli-0.6.5 → drydock_cli-0.6.7}/tests/test_loop_detection.py

@@ -59,18 +59,18 @@ class TestCircuitBreaker:
 
     def test_first_call_allowed(self):
         al = _make_agent()
-        tc = SimpleNamespace(tool_name="bash", raw_arguments='{"command":"ls -ltr"}')
+        tc = SimpleNamespace(tool_name="bash", args_dict={"command":"ls -ltr"})
         assert al._circuit_breaker_check(tc) is None
 
     def test_second_call_allowed(self):
         al = _make_agent()
-        tc = SimpleNamespace(tool_name="bash", raw_arguments='{"command":"ls -ltr"}')
+        tc = SimpleNamespace(tool_name="bash", args_dict={"command":"ls -ltr"})
         al._circuit_breaker_record(tc, "file1.py\nfile2.py")
         assert al._circuit_breaker_check(tc) is None
 
     def test_third_call_blocked(self):
         al = _make_agent()
-        tc = SimpleNamespace(tool_name="bash", raw_arguments='{"command":"ls -ltr"}')
+        tc = SimpleNamespace(tool_name="bash", args_dict={"command":"ls -ltr"})
         al._circuit_breaker_record(tc, "file1.py\nfile2.py")
         al._circuit_breaker_record(tc, "file1.py\nfile2.py")
         result = al._circuit_breaker_check(tc)
@@ -79,7 +79,7 @@ class TestCircuitBreaker:
 
     def test_blocked_message_contains_previous_result(self):
         al = _make_agent()
-        tc = SimpleNamespace(tool_name="bash", raw_arguments='{"command":"ls -ltr"}')
+        tc = SimpleNamespace(tool_name="bash", args_dict={"command":"ls -ltr"})
         al._circuit_breaker_record(tc, "important_file.py")
         al._circuit_breaker_record(tc, "important_file.py")
         result = al._circuit_breaker_check(tc)
@@ -87,7 +87,7 @@ class TestCircuitBreaker:
 
     def test_blocked_message_contains_alternatives(self):
         al = _make_agent()
-        tc = SimpleNamespace(tool_name="bash", raw_arguments='{"command":"ls -ltr"}')
+        tc = SimpleNamespace(tool_name="bash", args_dict={"command":"ls -ltr"})
         al._circuit_breaker_record(tc, "output")
         al._circuit_breaker_record(tc, "output")
         result = al._circuit_breaker_check(tc)
@@ -95,8 +95,8 @@ class TestCircuitBreaker:
 
     def test_different_args_not_blocked(self):
         al = _make_agent()
-        tc1 = SimpleNamespace(tool_name="bash", raw_arguments='{"command":"ls -ltr"}')
-        tc2 = SimpleNamespace(tool_name="bash", raw_arguments='{"command":"ls -la"}')
+        tc1 = SimpleNamespace(tool_name="bash", args_dict={"command":"ls -ltr"})
+        tc2 = SimpleNamespace(tool_name="bash", args_dict={"command":"ls -la"})
         al._circuit_breaker_record(tc1, "r1")
         al._circuit_breaker_record(tc1, "r1")
         assert al._circuit_breaker_check(tc1) is not None  # Blocked
@@ -104,8 +104,8 @@ class TestCircuitBreaker:
 
     def test_different_tools_not_blocked(self):
         al = _make_agent()
-        tc1 = SimpleNamespace(tool_name="bash", raw_arguments='{"command":"ls"}')
-        tc2 = SimpleNamespace(tool_name="grep", raw_arguments='{"pattern":"foo"}')
+        tc1 = SimpleNamespace(tool_name="bash", args_dict={"command":"ls"})
+        tc2 = SimpleNamespace(tool_name="grep", args_dict={"pattern":"foo"})
         al._circuit_breaker_record(tc1, "r1")
         al._circuit_breaker_record(tc1, "r1")
         assert al._circuit_breaker_check(tc1) is not None
@@ -113,7 +113,7 @@ class TestCircuitBreaker:
 
     def test_already_attempted_summary(self):
         al = _make_agent()
-        tc = SimpleNamespace(tool_name="bash", raw_arguments='{"command":"ls"}')
+        tc = SimpleNamespace(tool_name="bash", args_dict={"command":"ls"})
         al._circuit_breaker_record(tc, "output1")
         al._circuit_breaker_record(tc, "output1")
         result = al._circuit_breaker_check(tc)
@@ -327,7 +327,7 @@ class TestRealWorldLoops:
     def test_ls_ltr_loop(self):
         """User reports: `ls -ltr` runs 5+ times."""
         al = _make_agent()
-        tc = SimpleNamespace(tool_name="bash", raw_arguments='{"command":"ls -ltr"}')
+        tc = SimpleNamespace(tool_name="bash", args_dict={"command":"ls -ltr"})
 
         # First 2 calls go through
         al._circuit_breaker_record(tc, "total 48\n-rw-r--r-- 1 user user 1234 auth.py")