drydock-cli 0.7.1__tar.gz → 0.7.2__tar.gz

{drydock_cli-0.7.1 → drydock_cli-0.7.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: drydock-cli
-Version: 0.7.1
+Version: 0.7.2
 Summary: Drydock — nautical CLI coding agent
 Project-URL: Homepage, https://github.com/fbobe321/drydock
 Project-URL: Repository, https://github.com/fbobe321/drydock
@@ -168,6 +168,17 @@ export MISTRAL_API_KEY="your_key"            # Or set env var
 
 Keys are saved to `~/.drydock/.env`.
 
+### Consultant Model
+
+Set a smarter model for the `/consult` command:
+
+```toml
+# In ~/.drydock/config.toml
+consultant_model = "gemini-2.5-pro"
+```
+
+The consultant provides read-only advice — it never calls tools. Use `/consult <question>` to ask it. The response appears in your conversation so the local model can see it.
+
 ### Custom Agents
 
 Create agent configs in `~/.drydock/agents/`:

{drydock_cli-0.7.1 → drydock_cli-0.7.2}/README.md

@@ -114,6 +114,17 @@ export MISTRAL_API_KEY="your_key"            # Or set env var
 
 Keys are saved to `~/.drydock/.env`.
 
+### Consultant Model
+
+Set a smarter model for the `/consult` command:
+
+```toml
+# In ~/.drydock/config.toml
+consultant_model = "gemini-2.5-pro"
+```
+
+The consultant provides read-only advice — it never calls tools. Use `/consult <question>` to ask it. The response appears in your conversation so the local model can see it.
+
 ### Custom Agents
 
 Create agent configs in `~/.drydock/agents/`:

{drydock_cli-0.7.1 → drydock_cli-0.7.2}/drydock/cli/textual_ui/app.py

@@ -977,10 +977,18 @@ class VibeApp(App):  # noqa: PLR0904
 
         try:
             from drydock.core.consultant import ask_consultant
+            # Build conversation history for context
+            history_lines = []
+            for msg in self.agent_loop.messages[-10:]:  # Last 10 messages
+                if msg.role.value in ("user", "assistant") and msg.content:
+                    history_lines.append(f"[{msg.role.value}]: {msg.content[:200]}")
+            conversation_history = "\n".join(history_lines)
+
             advice = await ask_consultant(
                 question,
                 config=self.agent_loop.config,
                 model=consultant_model,
+                conversation_history=conversation_history,
             )
 
             if not advice or advice.startswith("(Consultant unavailable"):

{drydock_cli-0.7.1 → drydock_cli-0.7.2}/drydock/core/agent_loop.py

@@ -1518,7 +1518,7 @@ class AgentLoop:
 
         last_msg = self.messages[-1]
         if last_msg.role is Role.tool:
-            empty_assistant_msg = LLMMessage(role=Role.assistant, content="Understood.")
+            empty_assistant_msg = LLMMessage(role=Role.assistant, content="Continuing...")
             self.messages.append(empty_assistant_msg)
 
     def _reset_session(self) -> None:

{drydock_cli-0.7.1 → drydock_cli-0.7.2}/drydock/core/consultant.py

@@ -28,6 +28,7 @@ async def ask_consultant(
     question: str,
     config: object | None = None,
     model: str | None = None,
+    conversation_history: str = "",
 ) -> str:
     """Ask the consultant model a question using DryDock's backend.
 
@@ -52,10 +53,15 @@ async def ask_consultant(
     if not model_name:
         return ""
 
+    history_section = ""
+    if conversation_history:
+        history_section = f"\nCONVERSATION CONTEXT (recent messages):\n{conversation_history[:3000]}\n"
+
     prompt = (
         "You are a senior software engineering consultant. An AI coding agent is "
         "asking for your advice. Give concise, actionable guidance. Do NOT provide "
-        "code unless specifically asked — just say WHAT to do and WHERE to look.\n\n"
+        "code unless specifically asked — just say WHAT to do and WHERE to look.\n"
+        f"{history_section}\n"
         f"QUESTION:\n{question}"
     )
 

{drydock_cli-0.7.1 → drydock_cli-0.7.2}/drydock/core/prompts/cli.md

@@ -1,4 +1,4 @@
-You are Drydock, a CLI coding agent. You interact with a local codebase through tools. You have no internet access.
+You are DryDock, a CLI coding agent. You interact with a local codebase through tools. You have no internet access.
 CRITICAL: Users complain you are too verbose. Your responses must be minimal. Most tasks need <100 words. Code speaks for itself.
 
 Skills are markdown files in your skill directories, NOT tools or agents. To use a skill:

drydock_cli-0.7.2/drydock/skills/deep-research/SKILL.md

@@ -0,0 +1,57 @@
+---
+name: deep-research
+description: Deep research a topic using web search, code exploration, and multi-step analysis. Produces a structured report.
+user-invocable: true
+allowed-tools:
+  - bash
+  - grep
+  - read_file
+  - webfetch
+  - websearch
+  - write_file
+---
+
+# Deep Research
+
+Conduct thorough research on a topic and produce a structured report.
+
+## Workflow
+
+1. **Clarify scope**: What exactly needs to be researched? Ask if unclear.
+2. **Web search**: Use `websearch` for current information, articles, docs.
+3. **Code exploration**: Use `grep` and `read_file` to find relevant code patterns.
+4. **Cross-reference**: Compare web findings with codebase reality.
+5. **Write report**: Produce a structured markdown report with findings.
+
+## Report Template
+
+Write the report to `research_report.md`:
+
+```markdown
+# Research: [Topic]
+
+## Summary
+One paragraph overview of findings.
+
+## Key Findings
+- Finding 1 with source
+- Finding 2 with source
+- Finding 3 with source
+
+## Code Analysis
+Relevant code patterns found in the codebase.
+
+## Recommendations
+Actionable next steps based on research.
+
+## Sources
+- [Source 1](url)
+- [Source 2](url)
+```
+
+## Rules
+
+- Always cite sources (URLs for web, file:line for code)
+- If websearch fails (SSL/network), fall back to code-only analysis
+- Keep the report under 500 lines
+- Focus on actionable findings, not background

{drydock_cli-0.7.1 → drydock_cli-0.7.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "drydock-cli"
-version = "0.7.1"
+version = "0.7.2"
 description = "Drydock — nautical CLI coding agent"
 readme = "README.md"
 requires-python = ">=3.12"
@@ -76,6 +76,7 @@ include = ["drydock/"]
 
 [tool.hatch.build.targets.wheel.force-include]
 "drydock/skills/create-presentation" = "drydock/skills/create-presentation"
+"drydock/skills/deep-research" = "drydock/skills/deep-research"
 
 
 [project.scripts]

{drydock_cli-0.7.1 → drydock_cli-0.7.2}/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 tests/test_agent_tasks.py tests/test_integration.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 tests/test_integration.py tests/test_user_issues.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-0.7.1 → drydock_cli-0.7.2}/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 tests/test_agent_tasks.py tests/test_integration.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 tests/test_integration.py tests/test_user_issues.py \
     -p no:xdist -p no:cov --override-ini="addopts=" -q 2>&1; then
     log "TESTS FAILED — publish aborted."
     exit 1

drydock_cli-0.7.2/tests/test_user_issues.py

@@ -0,0 +1,259 @@
+"""Tests for user-reported issues.
+
+Each test should FAIL before the fix, proving the issue exists.
+After fixing, all should pass.
+"""
+
+from __future__ import annotations
+
+import inspect
+import os
+from pathlib import Path
+from types import SimpleNamespace
+from unittest.mock import patch
+
+import pytest
+
+
+# ============================================================================
+# Issue 1: "DryDock" branding — should be DryDock not Drydock everywhere
+# ============================================================================
+
+class TestBranding:
+    def test_readme_says_drydock_capital(self):
+        readme = Path(__file__).parent.parent / "README.md"
+        content = readme.read_text()
+        # Title should be "# DryDock" not "# Drydock"
+        assert "# DryDock" in content
+
+    def test_system_prompt_says_drydock(self):
+        prompt = Path(__file__).parent.parent / "drydock" / "core" / "prompts" / "cli.md"
+        content = prompt.read_text()
+        # Should say "DryDock" not "Drydock" as the product name
+        assert "DryDock" in content
+
+    def test_loading_widget_says_drydock(self):
+        from drydock.cli.textual_ui.widgets.loading import LoadingWidget
+        widget = LoadingWidget.__new__(LoadingWidget)
+        status = widget._get_default_status()
+        # Anchor emoji is fine, but product references should be DryDock
+        assert status  # Just verify it doesn't crash
+
+
+# ============================================================================
+# Issue 5: What conda env is pip using? Should preserve user's env
+# ============================================================================
+
+class TestCondaEnv:
+    def test_bash_env_preserves_active_conda(self):
+        """If user has CONDA_DEFAULT_ENV set, bash should preserve it."""
+        from drydock.core.tools.builtins.bash import _get_base_env
+        with patch.dict(os.environ, {"CONDA_DEFAULT_ENV": "myproject", "CONDA_EXE": "/opt/conda/bin/conda"}):
+            env = _get_base_env()
+            # BASH_ENV should reference the user's env, not just base
+            if "BASH_ENV" in env:
+                bash_env_content = Path(env["BASH_ENV"]).read_text() if Path(env["BASH_ENV"]).exists() else ""
+                # Should activate the user's env, not just source conda.sh
+                assert "myproject" in bash_env_content or "CONDA_DEFAULT_ENV" in str(env)
+
+
+# ============================================================================
+# Issue 6: Circuit breaker shows message but still repeats
+# ============================================================================
+
+class TestCircuitBreakerEnforcement:
+    def test_circuit_breaker_blocks_not_just_warns(self):
+        """After 2 identical calls, the 3rd must return an ERROR, not execute."""
+        from drydock.core.agent_loop import AgentLoop
+        from drydock.core.types import MessageList
+
+        al = object.__new__(AgentLoop)
+        al.messages = MessageList()
+        al._tool_call_history = {}
+
+        tc = SimpleNamespace(tool_name="bash", args_dict={"command": "ls -ltr"})
+        al._circuit_breaker_record(tc, "file1.py")
+        al._circuit_breaker_record(tc, "file1.py")
+
+        result = al._circuit_breaker_check(tc)
+        assert result is not None, "Circuit breaker should block"
+        assert "CIRCUIT BREAKER" in result
+        # The blocked message should tell the model to STOP, not just warn
+        assert "STOP" in result or "Do NOT" in result or "MUST" in result
+
+    def test_circuit_breaker_resets_for_new_conversation(self):
+        """Circuit breaker should reset between conversations, not cache forever."""
+        from drydock.core.agent_loop import AgentLoop
+        from drydock.core.types import MessageList
+
+        al = object.__new__(AgentLoop)
+        al.messages = MessageList()
+        al._tool_call_history = {}
+
+        tc = SimpleNamespace(tool_name="bash", args_dict={"command": "ls"})
+        al._circuit_breaker_record(tc, "r1")
+        al._circuit_breaker_record(tc, "r1")
+        assert al._circuit_breaker_check(tc) is not None  # Blocked
+
+        # Simulate new conversation — history should be clearable
+        al._tool_call_history.clear()
+        assert al._circuit_breaker_check(tc) is None  # Unblocked
+
+
+# ============================================================================
+# Issue 7: "Understood" after errors then stops
+# ============================================================================
+
+class TestUnderstoodBug:
+    def test_no_understood_in_agent(self):
+        """The agent should not inject 'Understood.' which causes premature stops."""
+        from drydock.core.agent_loop import AgentLoop
+        src = inspect.getsource(AgentLoop)
+        # "Understood." was causing the model to think conversation ended
+        assert "Understood" not in src, \
+            "'Understood' found in agent_loop — should use 'Continuing...' instead"
+
+
+# ============================================================================
+# Issue 10: Circuit breaker too sensitive — same command later should be OK
+# ============================================================================
+
+class TestCircuitBreakerSensitivity:
+    def test_circuit_breaker_should_have_decay(self):
+        """The circuit breaker should eventually allow re-running a command."""
+        from drydock.core.agent_loop import AgentLoop
+        from drydock.core.types import MessageList
+
+        al = object.__new__(AgentLoop)
+        al.messages = MessageList()
+        al._tool_call_history = {}
+
+        tc = SimpleNamespace(tool_name="bash", args_dict={"command": "git status"})
+        al._circuit_breaker_record(tc, "clean")
+        al._circuit_breaker_record(tc, "clean")
+        assert al._circuit_breaker_check(tc) is not None  # Blocked
+
+        # After clearing (simulating passage of time / new task), should work again
+        al._tool_call_history.clear()
+        assert al._circuit_breaker_check(tc) is None
+
+
+# ============================================================================
+# Issue 11: /consult needs to show which model is being used
+# ============================================================================
+
+class TestConsultModelDisplay:
+    def test_consult_handler_shows_model_name(self):
+        """The /consult response should include the model name."""
+        from drydock.cli.textual_ui.app import VibeApp
+        src = inspect.getsource(VibeApp._consult_command)
+        # Should display which model is being used
+        assert "consultant_model" in src or "model" in src
+
+    def test_consult_help_shows_config_instructions(self):
+        """Typing /consult with no args should show how to configure."""
+        from drydock.cli.textual_ui.app import VibeApp
+        src = inspect.getsource(VibeApp._consult_command)
+        assert "config.toml" in src or "consultant_model" in src
+
+
+# ============================================================================
+# Issue 12: SSL error on websearch — should use -k / DRYDOCK_INSECURE
+# ============================================================================
+
+class TestWebSearchSSL:
+    def test_websearch_respects_insecure_flag(self):
+        """WebSearch should check DRYDOCK_INSECURE env var."""
+        from drydock.core.tools.builtins.websearch import WebSearch
+        src = inspect.getsource(WebSearch)
+        assert "DRYDOCK_INSECURE" in src or "insecure" in src.lower() or "verify" in src
+
+    def test_webfetch_respects_insecure_flag(self):
+        """WebFetch should check DRYDOCK_INSECURE env var."""
+        from drydock.core.tools.builtins.webfetch import WebFetch
+        src = inspect.getsource(WebFetch)
+        assert "DRYDOCK_INSECURE" in src
+
+
+# ============================================================================
+# Issue 14: Config.toml consultant_model setup
+# ============================================================================
+
+class TestConsultantConfig:
+    def test_config_has_consultant_model_field(self):
+        """VibeConfig should have consultant_model field."""
+        from drydock.core.config import VibeConfig
+        assert hasattr(VibeConfig, "model_fields")
+        assert "consultant_model" in VibeConfig.model_fields
+
+    def test_consultant_model_documented_in_readme(self):
+        """README should mention how to set consultant_model."""
+        readme = Path(__file__).parent.parent / "README.md"
+        content = readme.read_text()
+        assert "consultant_model" in content or "consultant" in content.lower()
+
+
+# ============================================================================
+# Issue 15: Auto-consult agent — model should ask consultant when stuck
+# ============================================================================
+
+class TestAutoConsult:
+    def test_circuit_breaker_suggests_consult(self):
+        """When circuit breaker fires, it should mention /consult."""
+        from drydock.core.agent_loop import AgentLoop
+        from drydock.core.types import MessageList
+
+        al = object.__new__(AgentLoop)
+        al.messages = MessageList()
+        al._tool_call_history = {}
+
+        tc = SimpleNamespace(tool_name="bash", args_dict={"command": "ls"})
+        al._circuit_breaker_record(tc, "output")
+        al._circuit_breaker_record(tc, "output")
+        result = al._circuit_breaker_check(tc)
+        # Should suggest using /consult or mention consultant
+        assert "/consult" in result or "consultant" in result.lower() or "different" in result.lower()
+
+
+# ============================================================================
+# Issue 16: /consult should include conversation history
+# ============================================================================
+
+class TestConsultHistory:
+    def test_consult_sends_context(self):
+        """/consult handler should send conversation context to consultant."""
+        from drydock.core.consultant import ask_consultant
+        src = inspect.getsource(ask_consultant)
+        # Should accept and use context/conversation history
+        assert "context" in src or "messages" in src or "history" in src
+
+
+# ============================================================================
+# Issue 17: Deep research skill should exist
+# ============================================================================
+
+class TestDeepResearchSkill:
+    def test_research_skill_exists(self):
+        """A deep-research skill should be bundled."""
+        skills_dir = Path(__file__).parent.parent / "drydock" / "skills"
+        skill_dirs = [d.name for d in skills_dir.iterdir() if d.is_dir()] if skills_dir.exists() else []
+        has_research = any("research" in d for d in skill_dirs)
+        assert has_research, f"No research skill found. Available: {skill_dirs}"
+
+
+# ============================================================================
+# Issue 18: Not using agents — explore/diagnostic agents should be utilized
+# ============================================================================
+
+class TestAgentsAvailable:
+    def test_explore_agent_registered(self):
+        """The 'explore' agent should be registered."""
+        from drydock.core.agents.models import BuiltinAgentName
+        assert hasattr(BuiltinAgentName, "EXPLORE") or "explore" in dir(BuiltinAgentName)
+
+    def test_system_prompt_mentions_agents(self):
+        """System prompt should tell the model about available agents."""
+        prompt = Path(__file__).parent.parent / "drydock" / "core" / "prompts" / "cli.md"
+        content = prompt.read_text()
+        has_agent_ref = "agent" in content.lower() or "subagent" in content.lower() or "task" in content.lower()
+        assert has_agent_ref