slack-claude-code 0.1.5__py3-none-any.whl

slack_claude_code-0.1.5.dist-info/METADATA

@@ -0,0 +1,226 @@
+Metadata-Version: 2.4
+Name: slack-claude-code
+Version: 0.1.5
+Summary: Slack app for running Claude Code CLI commands
+License-File: LICENSE
+Author: Dan
+Requires-Python: >=3.10,<4.0
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
+Requires-Dist: aiofiles (>=24.1.0,<25.0.0)
+Requires-Dist: aiohttp (>=3.13.2,<4.0.0)
+Requires-Dist: aiosqlite (>=0.21.0,<0.22.0)
+Requires-Dist: cryptography (>=44.0.0,<45.0.0)
+Requires-Dist: loguru (>=0.7.0,<0.8.0)
+Requires-Dist: pydantic-settings (>=2.7.0,<3.0.0)
+Requires-Dist: slack-bolt (>=1.27.0,<2.0.0)
+Description-Content-Type: text/markdown
+
+<p align="center">
+  <img src="assets/repo_logo.png" alt="Slack Claude Code Bot" width="1000">
+</p>
+
+<p align="center">
+  <a href="https://pypi.org/project/slack-claude-code/"><img src="https://img.shields.io/pypi/v/slack-claude-code" alt="PyPI version"></a>
+  <a href="https://pypi.org/project/slack-claude-code/"><img src="https://img.shields.io/pypi/pyversions/slack-claude-code" alt="Python versions"></a>
+  <a href="https://opensource.org/licenses/MIT"><img src="https://img.shields.io/badge/license-MIT-blue.svg" alt="License"></a>
+  <a href="https://github.com/djkelleher/slack-claude-code/actions/workflows/tests.yml"><img src="https://github.com/djkelleher/slack-claude-code/actions/workflows/tests.yml/badge.svg" alt="Tests"></a>
+</p>
+
+**Claude Code, but in Slack.** Access Claude Code remotely from any device, or use it full-time for a better UI experience.
+
+## Why Slack?
+
+| Feature | Terminal | Slack |
+|---------|----------|-------|
+| **Code blocks** | Plain text | Syntax-highlighted with copy button |
+| **Long output** | Scrolls off screen | "View Details" modal |
+| **Permissions** | Y/n prompts | Approve/Deny buttons |
+| **Parallel work** | Multiple terminals | Threads = isolated sessions |
+| **File sharing** | `cat` or copy-paste | Drag & drop with preview |
+| **Notifications** | Watch the terminal | Alerts when tasks complete |
+| **Streaming** | Live terminal output | Watch responses as they generate |
+| **Smart context** | Manual file inclusion | Frequently-used files auto-included |
+
+## Installation
+
+### Prerequisites
+- Python 3.10+
+- [Claude Code CLI](https://github.com/anthropics/claude-code) installed and authenticated
+
+### 1. Install the `ccslack` executable
+```bash
+pipx install slack-claude-code
+```
+
+### 2. Create Slack App
+Go to https://api.slack.com/apps → "Create New App" → "From scratch"
+
+**Socket Mode**: Enable and create an app-level token with `connections:write` scope (save the `xapp-` token)
+
+**Bot Token Scopes** (OAuth & Permissions):
+- `chat:write`, `commands`, `channels:history`, `app_mentions:read`, `files:read`, `files:write`
+
+**Event Subscriptions**: Enable and add `message.channels`, `app_mention`
+
+**App Icon**: In "Basic Information" → "Display Information", upload `assets/claude_logo.png` from this repo as the app icon
+
+**Slash Commands**: Add the commands from the tables below (or the subset that you plan to use)
+
+#### Configuration
+Customize Claude's behavior for your workflow.
+
+| Command | Description | Example |
+|---------|-------------|---------|
+| `/model` | Show or change AI model | `/model sonnet` |
+| `/mode` | View or set permission mode | `/mode`, `/mode plan` |
+| `/permissions` | View/update permissions | `/permissions` |
+| `/notifications` | Configure notifications | `/notifications` |
+
+#### Session Management
+Each Slack thread maintains an isolated Claude session with its own context.
+
+| Command | Description | Example |
+|---------|-------------|---------|
+| `/clear` | Reset conversation | `/clear` |
+| `/compact` | Compact context | `/compact` |
+| `/cost` | Show session cost | `/cost` |
+| `/resume` | Resume previous session | `/resume` |
+| `/sessions` | List active sessions | `/sessions` |
+| `/session-cleanup` | Clean up inactive sessions | `/session-cleanup` |
+
+#### Navigation
+Control the working directory for Claude's file operations.
+
+| Command | Description | Example |
+|---------|-------------|---------|
+| `/ls` | List directory contents | `/ls`, `/ls src/` |
+| `/cd` | Change working directory | `/cd /home/user/project`, `cd subfolder`, `cd ..` |
+| `/pwd` | Print working directory | `/pwd` |
+| `/add-dir` | Add directory to context | `/add-dir ./lib` |
+
+#### CLI Tools
+Direct access to Claude Code CLI functionality.
+
+| Command | Description | Example |
+|---------|-------------|---------|
+| `/init` | Initialize project config | `/init` |
+| `/memory` | View/edit Claude's memory | `/memory` |
+| `/review` | Review code changes | `/review` |
+| `/doctor` | Diagnose installation | `/doctor` |
+| `/stats` | Show session statistics | `/stats` |
+| `/context` | Display context info | `/context` |
+| `/todos` | List and manage todos | `/todos` |
+| `/claude-help` | Show Claude Code help | `/claude-help` |
+| `/claude-config` | Show configuration | `/claude-config` |
+
+#### Multi-Agent Tasks
+Autonomous Planner → Worker → Evaluator pipeline for complex tasks. Iterates up to 3 times until complete.
+
+| Command | Description | Example |
+|---------|-------------|---------|
+| `/task` | Start a multi-agent task | `/task add unit tests for UserService` |
+| `/tasks` | List active tasks with status | `/tasks` |
+| `/task-cancel` | Cancel a running task | `/task-cancel abc123` |
+
+#### Command Queue
+Queue commands for sequential execution while preserving Claude's session context across items.
+
+| Command | Description | Example |
+|---------|-------------|---------|
+| `/q` | Add command to queue | `/q analyze the API endpoints` |
+| `/qv` | View queue status | `/qv` |
+| `/qc` | Clear pending queue | `/qc` |
+| `/qr` | Remove specific item | `/qr 5` |
+
+#### Jobs & Control
+Monitor and control long-running operations with real-time progress updates.
+
+| Command | Description | Example |
+|---------|-------------|---------|
+| `/st` | Show active job status | `/st` |
+| `/cc` | Cancel jobs | `/cc` or `/cc abc123` |
+| `/esc` | Send interrupt (Ctrl+C) | `/esc` |
+
+#### Git
+Full git workflow without leaving Slack. Includes branch name and commit message validation.
+
+| Command | Description | Example |
+|---------|-------------|---------|
+| `/status` | Show branch and changes | `/status` |
+| `/diff` | Show uncommitted changes | `/diff --staged` |
+| `/commit` | Commit staged changes | `/commit fix: resolve race condition` |
+| `/branch` | Manage branches | `/branch create feature/auth` |
+
+
+### 3. Configure
+
+Use the built-in config CLI to securely store your Slack credentials:
+
+```bash
+ccslack-config set SLACK_BOT_TOKEN=xoxb-...
+ccslack-config set SLACK_APP_TOKEN=xapp-...
+ccslack-config set SLACK_SIGNING_SECRET=...
+```
+
+**Config CLI Commands:**
+| Command | Description |
+|---------|-------------|
+| `ccslack-config set KEY=VALUE` | Store a configuration value |
+| `ccslack-config get KEY` | Retrieve a configuration value |
+| `ccslack-config list` | List all stored configuration |
+| `ccslack-config delete KEY` | Remove a configuration value |
+| `ccslack-config path` | Show config file locations |
+
+Configuration is encrypted and stored in `~/.slack-claude-code/config.enc`. Sensitive values (tokens, secrets) are masked when displayed.
+
+**Alternative:** You can also use environment variables or a `.env` file. Config values take precedence over environment variables.
+
+**Where to find these values:**
+- `SLACK_BOT_TOKEN`: Your App → OAuth & Permissions → Bot User OAuth Token
+- `SLACK_APP_TOKEN`: Your App → Basic Information → App-Level Tokens → (token you created with `connections:write`)
+- `SLACK_SIGNING_SECRET`: Your App → Basic Information → App Credentials → Signing Secret
+
+### 4. Start the Slack bot
+You can now run `ccslack` in your terminal. The working directory where you start the executable will be the default working directory for your Claude Code session(s). If you have a .env file in this directory, it will automatically be loaded.   
+
+## Usage
+
+Type messages in any channel where the bot is installed. The main channel is a single Claude Code session. If you click `reply` to any message and start a thread, this will be a new Claude Code session.
+
+## Architecture
+
+```
+src/
+├── app.py                 # Main entry point
+├── config.py              # Configuration
+├── database/              # SQLite persistence (models, migrations, repository)
+├── claude/                # Claude CLI integration (streaming)
+├── handlers/              # Slack command handlers
+├── agents/                # Multi-agent orchestration (planner→worker→evaluator)
+├── approval/              # Permission & plan approval handling
+├── git/                   # Git operations (status, diff, commit, branch)
+├── hooks/                 # Event hook system
+├── question/              # AskUserQuestion tool support
+├── tasks/                 # Background task management
+└── utils/                 # Formatters, helpers, validators
+```
+
+## Troubleshooting
+
+| Problem | Solution |
+|---------|----------|
+| Configuration errors on startup | Check `.env` has all required tokens |
+| Commands not appearing | Verify slash commands in Slack app settings |
+
+## License
+
+MIT
+
+- - - 
+
+Congratulations, you can now use Claude Code from anywhere 🎉💪

slack_claude_code-0.1.5.dist-info/RECORD

@@ -0,0 +1,65 @@
+src/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
+src/agents/__init__.py,sha256=XajhwXQX2DmxDoovXa7tkf0TnQ7tgEfNf5PDA3WuOaU,210
+src/agents/orchestrator.py,sha256=8Xi_K_NuOWt8q4fh_OoRm3suntqAySYysj4oXSL8dyQ,10187
+src/agents/roles.py,sha256=nTsO1wTjrkpiLYcXweX0KJr_s1kIeB9NRTYzfX05t6w,4464
+src/app.py,sha256=f69KgfMb7qq9MiER2vbMVV9xvKAOJ1fQS4RrzS-66LE,32038
+src/approval/__init__.py,sha256=39OldyhgimZqGuB_RMrc9cRafpqeU497w-bG0dWJMRI,46
+src/approval/handler.py,sha256=KEN50hB4IkE6urhJ2JaYPweAnvtceUx7JlzFvvnyPkQ,9444
+src/approval/plan_manager.py,sha256=PVhyqWiphvr72aKlb8tHFDZBcEx8Svka2ICk7ceBqAg,6754
+src/approval/slack_ui.py,sha256=rcrSTCytJIdWoT8unPTKuufgJbcy52HqkMmjobXxr4g,7671
+src/claude/__init__.py,sha256=lVuqdU_IGGwuuPrxpB5hZU32MPrqLzd5wxqdu_uJHbY,103
+src/claude/streaming.py,sha256=FvlEt1zWKG_Rti4jl5-wdkCByaEZmvw899fipxAdqGA,14166
+src/claude/subprocess_executor.py,sha256=kG7d7oy_7f1qjJp9-TeJ6fZhHANnf3zOmJgmmVR9YI8,21730
+src/cli.py,sha256=SYaTHNk3p9KTdqxeLgYZTW_N--yLbtNpjxZFbsHgvoQ,4642
+src/config.py,sha256=dpJMdn3e-MBcxb8mERkQk4JUym5loX6LkHh5n8OOx0Y,7305
+src/config_storage.py,sha256=zttA1u-lmKNjy424v2j75s-TCZOBiwSrBh6XG5i8U_s,4346
+src/database/__init__.py,sha256=MAJqxqwvti69iZCcBRHiuN42tUvDI10rXO2r-ms5gp4,45
+src/database/migrations.py,sha256=JZCKiNXXGkalCLVnD4Ml_gsASe005jCdD0592XCvtsI,6773
+src/database/models.py,sha256=TltB9Yhsdc7w-W9fLb9vzu-wUnUhb7pEw2t5e5LNoY4,9150
+src/database/repository.py,sha256=nNt3j1sy-HupszaLQ2Hez3FP33qaQMljildBJ_OQqHM,30288
+src/git/__init__.py,sha256=cV5jpkmZRlpj4eBLHvwliA3Bc1dyEb65K8la9VqEjew,54
+src/git/models.py,sha256=hRdHffabRuZuRkgyTJOJe8ridqC5mErSvmlO91VWDTY,1718
+src/git/service.py,sha256=JS75Q2R-D_WTnLNgLahDYXX4lyrfhi24DOAc-5KnPoc,12394
+src/handlers/__init__.py,sha256=c3f34SBl1jDU2n1W0al-WcqSnrCTeESOTqCOaKIAhnY,1646
+src/handlers/actions.py,sha256=A3xareMrWw-FPLblHlbQ-_PjbVfxAiY7zCLqklM8UNM,37022
+src/handlers/agents.py,sha256=AFpj_Sa4IXwU3oMkwNts-w504UxZ0niQgXnnkHEFBIk,10217
+src/handlers/base.py,sha256=HFYxdL-vJUbgTf_C5FwlgzLIHTOGVOGwh_bAerdfVXU,5045
+src/handlers/basic.py,sha256=mm94kELgdcwOwtk3ehlP2ETUpmQKnzWbfLjoEBEkkV0,5393
+src/handlers/claude_cli.py,sha256=T3AK42hfr5fl9ydmr288P71wXp0y93ooL1REqUSu0II,14631
+src/handlers/git.py,sha256=KZDkUrd3aksz6huQdcsO8r598yg7CnOmh0ABYX1GZwo,12490
+src/handlers/mode.py,sha256=lMDu2s5WSjJftv0nd46vRM2ent4cgF2C31mTTZ48LIE,4167
+src/handlers/notifications.py,sha256=dXHI8Sy4wPq2nHuhXBt7uiPyJZKf3jnmF_F0WJ2dAlM,7523
+src/handlers/parallel.py,sha256=EnYUdoZBcoA2ZqfE2ng2i7xyJmYJS45ds3EQCn2cGnk,2606
+src/handlers/queue.py,sha256=g8gTp88J1Idc6BMFgyDV-TGq_4IC1drKOuUQrTYa9Fs,9578
+src/handlers/session_management.py,sha256=bl4_98G6_ttWJlCQ68g8COJosBbuJ4LERyGMsY9yBzI,1730
+src/hooks/__init__.py,sha256=RH90banbNzw0ew-bkKdKV5ztlOotObMmz-t21Hj6zl4,165
+src/hooks/registry.py,sha256=WDzY44bOLSJYQ-CWFIoGujps7TR4xWPTLTL8G4dUIcU,6205
+src/hooks/types.py,sha256=Ii5tf-aQnCyBoz2B0fLBlrt_KojfALVeSOrtS-ASJ2U,1404
+src/question/__init__.py,sha256=qRO6m9GTxepXekPF6WoNtDDqyQmG5U5FCyqs9zI4XNo,59
+src/question/manager.py,sha256=sauP_PIk8lnKyTcvHC2hQQwFgBB_qrNiFzgZlLFREjE,13485
+src/question/slack_ui.py,sha256=cFupRThaGCUEtYkTqISJAaWpFVfnh9eRYWZPg0le-O8,9088
+src/tasks/__init__.py,sha256=mNU8fdvI4Ag3iQhewSLIF9oo_-m1Wtcimfw18M8d6fY,33
+src/tasks/manager.py,sha256=1j1uoEfazVEhER3oRByfAtHmq4efkpavcnO3CmpCypw,10206
+src/utils/__init__.py,sha256=B-ohn1ey6nbzxmoe7OrAmxeWC5TdVoU00YJIb1YSbjg,50
+src/utils/detail_cache.py,sha256=sq1nIsgQUgHuF5AW82lu8p-tZwlHCZZiMJQR5kRI37k,2233
+src/utils/file_downloader.py,sha256=u2lOps_JCJV3K381J7DiUS-0vj6YJGuALiUJMthoIqE,4923
+src/utils/formatters/__init__.py,sha256=8dnaLxUxTs1Sv138r5BUavSA-tj580vO5T7rc-9g0qA,64
+src/utils/formatters/base.py,sha256=N5X0wPOOhO_BBHCP8-TMIEKv-iLoOheRSdmLve0vlXM,5833
+src/utils/formatters/command.py,sha256=YaKIvbblLvL6otySJwAskRt-8khha5NE2ov4ZobsPjY,4081
+src/utils/formatters/directory.py,sha256=anYs8uS6Md5udHAzm7pV2CrgkJOVt_n1N3IekgZBCqI,1672
+src/utils/formatters/job.py,sha256=vfrEu-JdEvLLMrQxNPbfkqb-etld2_3OdeewEatsOdc,5838
+src/utils/formatters/markdown.py,sha256=xQqs7RBtdufWkf7YwpeNA4Ph13RBcD1irsy7Bu8c5vQ,2870
+src/utils/formatters/plan.py,sha256=ijVkghHrxAKKtf04op2N1dN2ayvndf6xEDPGBM9gWns,4723
+src/utils/formatters/queue.py,sha256=ObRLx5oWEUfpJw2lUZru2XY05lLTdP5M2sqpx-2Gx0w,2987
+src/utils/formatters/session.py,sha256=o37ZX-_W2NVCT1VNUi01YOhgJPB0QgPyz9rsyY3W9-I,4158
+src/utils/formatters/streaming.py,sha256=HwX-xIsgXTr8tpWCHQA2ZsErFoutN-E_S2tNME5nZQQ,2508
+src/utils/formatters/tool_blocks.py,sha256=QY7rdo7dek9uFMDYDyZgHvCo0QOaYJL7J3siPMcl2f0,9582
+src/utils/formatting.py,sha256=IKT-r8JaFnrQwWA4zW8UWMBj1Za0BVTha2E7L3uexb8,4445
+src/utils/slack_helpers.py,sha256=U9P-YJ67RYROOBGsC9FdrdCCR4rojAAJ5RKrJlCL82k,9466
+src/utils/streaming.py,sha256=ypP_jyD-9hOXfkMZFspVXkacDq0P3JKAMuBqxiuZzs4,9359
+src/utils/validators.py,sha256=dOwr5WITYpANipH5uIOl6F7dBueUS3vPKqRNFcxhE-I,1075
+slack_claude_code-0.1.5.dist-info/METADATA,sha256=5F5j4LyBkveOGI2_0J6jedImq_aWRIiyeJhpXZnvcsU,9487
+slack_claude_code-0.1.5.dist-info/WHEEL,sha256=3ny-bZhpXrU6vSQ1UPG34FoxZBp3lVcvK0LkgUz6VLk,88
+slack_claude_code-0.1.5.dist-info/entry_points.txt,sha256=feO-FYg1CvU17Cu4PrUiZfgm3xzlRy3-j4RWAI8hReY,66
+slack_claude_code-0.1.5.dist-info/licenses/LICENSE,sha256=z_UgzTmrZGQju-Lm9z-Ge8BZHioEvlEUh-doyIVN5mE,1060
+slack_claude_code-0.1.5.dist-info/RECORD,,

slack_claude_code-0.1.5.dist-info/WHEEL

@@ -0,0 +1,4 @@
+Wheel-Version: 1.0
+Generator: poetry-core 2.3.0
+Root-Is-Purelib: true
+Tag: py3-none-any

slack_claude_code-0.1.5.dist-info/entry_points.txt

@@ -0,0 +1,4 @@
+[console_scripts]
+ccslack=src.app:run
+ccslack-config=src.cli:run
+

slack_claude_code-0.1.5.dist-info/licenses/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2025 Dan
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

src/agents/__init__.py

@@ -0,0 +1,10 @@
+"""Multi-agent workflow orchestration."""
+
+from .orchestrator import (
+    AgentTask,
+    EvalResult,
+    MultiAgentOrchestrator,
+    TaskStatus,
+    WorkflowResult,
+)
+from .roles import AgentConfig, AgentRole

src/agents/orchestrator.py

@@ -0,0 +1,315 @@
+"""Multi-agent workflow orchestrator.
+
+Coordinates Planner -> Worker -> Evaluator pipeline for complex tasks.
+"""
+
+import asyncio
+from dataclasses import dataclass, field
+from datetime import datetime
+from enum import Enum
+from typing import Awaitable, Callable, Optional
+
+from loguru import logger
+
+from ..claude.subprocess_executor import ExecutionResult, SubprocessExecutor
+from .roles import AgentRole, format_task_prompt
+
+
+class TaskStatus(Enum):
+    """Status of an agent task."""
+
+    PENDING = "pending"
+    PLANNING = "planning"
+    WORKING = "working"
+    EVALUATING = "evaluating"
+    COMPLETED = "completed"
+    FAILED = "failed"
+    CANCELLED = "cancelled"
+
+
+class EvalResult(Enum):
+    """Evaluation result from evaluator agent."""
+
+    COMPLETE = "complete"
+    PARTIAL = "partial"
+    INCOMPLETE = "incomplete"
+    FAILED = "failed"
+
+
+@dataclass
+class AgentTask:
+    """A task being processed by the multi-agent workflow."""
+
+    task_id: str
+    description: str
+    channel_id: str
+    working_directory: str = "~"
+    status: TaskStatus = TaskStatus.PENDING
+    plan_output: Optional[str] = None
+    work_output: Optional[str] = None
+    eval_output: Optional[str] = None
+    eval_result: Optional[EvalResult] = None
+    error_message: Optional[str] = None
+    turn_count: int = 0
+    max_turns: int = 50
+    slack_thread_ts: Optional[str] = None
+    message_ts: Optional[str] = None
+    created_at: datetime = field(default_factory=datetime.now)
+    started_at: Optional[datetime] = None
+    completed_at: Optional[datetime] = None
+
+
+@dataclass
+class WorkflowResult:
+    """Result of a complete workflow execution."""
+
+    task: AgentTask
+    success: bool
+    plan: Optional[str] = None
+    work_output: Optional[str] = None
+    evaluation: Optional[str] = None
+    eval_result: Optional[EvalResult] = None
+    total_turns: int = 0
+    duration_ms: Optional[int] = None
+
+
+class MultiAgentOrchestrator:
+    """Orchestrates multi-agent workflows.
+
+    Runs tasks through Planner -> Worker -> Evaluator pipeline.
+    Each agent uses its own isolated subprocess session.
+    """
+
+    def __init__(
+        self,
+        executor: SubprocessExecutor,
+        max_iterations: int = 3,
+    ) -> None:
+        """Initialize orchestrator.
+
+        Args:
+            executor: Claude executor for running agents
+            max_iterations: Max worker-evaluator iterations
+        """
+        self.executor = executor
+        self.max_iterations = max_iterations
+        self._active_tasks: dict[str, AgentTask] = {}
+
+    async def execute_workflow(
+        self,
+        task: AgentTask,
+        on_status_update: Optional[Callable[[AgentTask], Awaitable[None]]] = None,
+    ) -> WorkflowResult:
+        """Execute the full multi-agent workflow.
+
+        Args:
+            task: The task to execute
+            on_status_update: Optional callback for status updates
+
+        Returns:
+            WorkflowResult with outcomes
+        """
+        self._active_tasks[task.task_id] = task
+        task.started_at = datetime.now()
+        start_time = asyncio.get_running_loop().time()
+
+        try:
+            # Phase 1: Planning
+            task.status = TaskStatus.PLANNING
+            if on_status_update:
+                await on_status_update(task)
+
+            plan_result = await self._run_planner(task)
+            if not plan_result.success:
+                task.status = TaskStatus.FAILED
+                task.error_message = plan_result.error
+                return self._create_result(task, False, start_time)
+
+            task.plan_output = plan_result.output
+            task.turn_count += 1
+
+            # Phase 2: Working (with potential re-iterations)
+            for iteration in range(self.max_iterations):
+                task.status = TaskStatus.WORKING
+                if on_status_update:
+                    await on_status_update(task)
+
+                work_result = await self._run_worker(task)
+                if not work_result.success:
+                    task.status = TaskStatus.FAILED
+                    task.error_message = work_result.error
+                    return self._create_result(task, False, start_time)
+
+                task.work_output = work_result.output
+                task.turn_count += 1
+
+                # Phase 3: Evaluation
+                task.status = TaskStatus.EVALUATING
+                if on_status_update:
+                    await on_status_update(task)
+
+                eval_result = await self._run_evaluator(task)
+                task.eval_output = eval_result.output
+                task.turn_count += 1
+
+                # Parse evaluation result
+                eval_verdict = self._parse_eval_result(eval_result.output)
+                task.eval_result = eval_verdict
+
+                if eval_verdict == EvalResult.COMPLETE:
+                    task.status = TaskStatus.COMPLETED
+                    task.completed_at = datetime.now()
+                    return self._create_result(task, True, start_time)
+
+                elif eval_verdict == EvalResult.FAILED:
+                    task.status = TaskStatus.FAILED
+                    task.error_message = "Evaluation returned FAILED"
+                    return self._create_result(task, False, start_time)
+
+                # For PARTIAL or INCOMPLETE, continue to next iteration
+                logger.info(
+                    f"Task {task.task_id} eval: {eval_verdict.value}, "
+                    f"iteration {iteration + 1}/{self.max_iterations}"
+                )
+
+            # Max iterations reached
+            task.status = TaskStatus.COMPLETED
+            task.completed_at = datetime.now()
+            return self._create_result(task, True, start_time)
+
+        except asyncio.CancelledError:
+            task.status = TaskStatus.CANCELLED
+            raise
+
+        except Exception as e:
+            task.status = TaskStatus.FAILED
+            task.error_message = str(e)
+            logger.error(f"Workflow failed for task {task.task_id}: {e}")
+            return self._create_result(task, False, start_time)
+
+        finally:
+            self._active_tasks.pop(task.task_id, None)
+            # Cleanup agent sessions
+            await self._cleanup_sessions(task.task_id)
+
+    async def _run_planner(self, task: AgentTask) -> ExecutionResult:
+        """Run the planner agent."""
+        prompt = format_task_prompt(
+            role=AgentRole.PLANNER,
+            task=task.description,
+            working_directory=task.working_directory,
+        )
+
+        # Use task-specific session for planner
+        session_id = f"{task.task_id}-planner"
+
+        return await self.executor.execute(
+            prompt=prompt,
+            working_directory=task.working_directory,
+            session_id=session_id,
+            execution_id=session_id,
+        )
+
+    async def _run_worker(self, task: AgentTask) -> ExecutionResult:
+        """Run the worker agent."""
+        prompt = format_task_prompt(
+            role=AgentRole.WORKER,
+            task=task.description,
+            plan=task.plan_output,
+        )
+
+        # Use task-specific session for worker
+        session_id = f"{task.task_id}-worker"
+
+        return await self.executor.execute(
+            prompt=prompt,
+            working_directory=task.working_directory,
+            session_id=session_id,
+            execution_id=session_id,
+        )
+
+    async def _run_evaluator(self, task: AgentTask) -> ExecutionResult:
+        """Run the evaluator agent."""
+        prompt = format_task_prompt(
+            role=AgentRole.EVALUATOR,
+            task=task.description,
+            plan=task.plan_output,
+            work_output=task.work_output,
+        )
+
+        # Use task-specific session for evaluator
+        session_id = f"{task.task_id}-evaluator"
+
+        return await self.executor.execute(
+            prompt=prompt,
+            working_directory=task.working_directory,
+            session_id=session_id,
+            execution_id=session_id,
+        )
+
+    def _parse_eval_result(self, output: str) -> EvalResult:
+        """Parse evaluation verdict from output."""
+        output_upper = output.upper()
+
+        if "COMPLETE" in output_upper and "INCOMPLETE" not in output_upper:
+            return EvalResult.COMPLETE
+        elif "INCOMPLETE" in output_upper:
+            return EvalResult.INCOMPLETE
+        elif "PARTIAL" in output_upper:
+            return EvalResult.PARTIAL
+        elif "FAILED" in output_upper or "FAIL" in output_upper:
+            return EvalResult.FAILED
+
+        # Default to partial if can't determine
+        return EvalResult.PARTIAL
+
+    def _create_result(
+        self,
+        task: AgentTask,
+        success: bool,
+        start_time: float,
+    ) -> WorkflowResult:
+        """Create a workflow result."""
+        duration_ms = int((asyncio.get_running_loop().time() - start_time) * 1000)
+
+        return WorkflowResult(
+            task=task,
+            success=success,
+            plan=task.plan_output,
+            work_output=task.work_output,
+            evaluation=task.eval_output,
+            eval_result=task.eval_result,
+            total_turns=task.turn_count,
+            duration_ms=duration_ms,
+        )
+
+    async def _cleanup_sessions(self, task_id: str) -> None:
+        """Cancel any active subprocess executions for a task."""
+        for role in ["planner", "worker", "evaluator"]:
+            session_id = f"{task_id}-{role}"
+            await self.executor.cancel(session_id)
+
+    async def cancel_task(self, task_id: str) -> bool:
+        """Cancel an active task.
+
+        Args:
+            task_id: The task ID to cancel
+
+        Returns:
+            True if task was found and cancelled
+        """
+        task = self._active_tasks.get(task_id)
+        if not task:
+            return False
+
+        task.status = TaskStatus.CANCELLED
+        await self._cleanup_sessions(task_id)
+        return True
+
+    def get_active_tasks(self) -> list[AgentTask]:
+        """Get list of active tasks."""
+        return list(self._active_tasks.values())
+
+    def get_task(self, task_id: str) -> Optional[AgentTask]:
+        """Get a specific task by ID."""
+        return self._active_tasks.get(task_id)