ros/autocoder
1
0
Fork 0
mirror of https://github.com/leonvanzyl/autocoder.git synced 2026-01-30 06:12:06 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat: add dedicated testing agents and enhanced parallel orchestration

Browse Source 
Introduce a new testing agent architecture that runs regression tests
independently from coding agents, improving quality assurance in
parallel mode.

Key changes:

Testing Agent System:
- Add testing_prompt.template.md for dedicated testing agent role
- Add feature_mark_failing MCP tool for regression detection
- Add --agent-type flag to select initializer/coding/testing mode
- Remove regression testing from coding prompt (now handled by testing agents)

Parallel Orchestrator Enhancements:
- Add testing agent spawning with configurable ratio (--testing-agent-ratio)
- Add comprehensive debug logging system (DebugLog class)
- Improve database session management to prevent stale reads
- Add engine.dispose() calls to refresh connections after subprocess commits
- Fix f-string linting issues (remove unnecessary f-prefixes)

UI Improvements:
- Add testing agent mascot (Chip) to AgentAvatar
- Enhance AgentCard to display testing agent status
- Add testing agent ratio slider in SettingsModal
- Update WebSocket handling for testing agent updates
- Improve ActivityFeed to show testing agent activity

API & Server Updates:
- Add testing_agent_ratio to settings schema and endpoints
- Update process manager to support testing agent type
- Enhance WebSocket messages for agent_update events

Template Changes:
- Delete coding_prompt_yolo.template.md (consolidated into main prompt)
- Update initializer_prompt.template.md with improved structure
- Streamline coding_prompt.template.md workflow

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
Auto
2026-01-18 13:49:50 +02:00
parent 5f786078fa
commit 13128361b0
27 changed files with 1885 additions and 536 deletions
Download Patch File Download Diff File Expand all files Collapse all files

130
autonomous_agent_demo.py
View File

@@ -4,8 +4,10 @@ Autonomous Coding Agent Demo
============================
A minimal harness demonstrating long-running autonomous coding with Claude.
This script implements the two-agent pattern (initializer + coding agent) and
incorporates all the strategies from the long-running agents guide.
This script implements a unified orchestrator pattern that handles:
- Initialization (creating features from app_spec)
- Coding agents (implementing features)
- Testing agents (regression testing)
Example Usage:
# Using absolute path directly
@@ -14,17 +16,22 @@ Example Usage:
# Using registered project name (looked up from registry)
python autonomous_agent_demo.py --project-dir my-app
# Limit iterations for testing
# Limit iterations for testing (when running as subprocess)
python autonomous_agent_demo.py --project-dir my-app --max-iterations 5
# YOLO mode: rapid prototyping without browser testing
# YOLO mode: rapid prototyping without testing agents
python autonomous_agent_demo.py --project-dir my-app --yolo
# Parallel execution with 3 concurrent agents (default)
python autonomous_agent_demo.py --project-dir my-app --parallel
# Parallel execution with 3 concurrent coding agents
python autonomous_agent_demo.py --project-dir my-app --concurrency 3
# Parallel execution with 5 concurrent agents
python autonomous_agent_demo.py --project-dir my-app --parallel 5
# Single agent mode (orchestrator with concurrency=1, the default)
python autonomous_agent_demo.py --project-dir my-app
# Run as specific agent type (used by orchestrator to spawn subprocesses)
python autonomous_agent_demo.py --project-dir my-app --agent-type initializer
python autonomous_agent_demo.py --project-dir my-app --agent-type coding --feature-id 42
python autonomous_agent_demo.py --project-dir my-app --agent-type testing
"""
import argparse
@@ -44,25 +51,28 @@ from registry import DEFAULT_MODEL, get_project_path
def parse_args() -> argparse.Namespace:
"""Parse command line arguments."""
parser = argparse.ArgumentParser(
description="Autonomous Coding Agent Demo - Long-running agent harness",
description="Autonomous Coding Agent Demo - Unified orchestrator pattern",
formatter_class=argparse.RawDescriptionHelpFormatter,
epilog="""
Examples:
# Use absolute path directly
# Use absolute path directly (single agent, default)
python autonomous_agent_demo.py --project-dir C:/Projects/my-app
# Use registered project name (looked up from registry)
python autonomous_agent_demo.py --project-dir my-app
# Use a specific model
python autonomous_agent_demo.py --project-dir my-app --model claude-sonnet-4-5-20250929
# Parallel execution with 3 concurrent agents
python autonomous_agent_demo.py --project-dir my-app --concurrency 3
# Limit iterations for testing
python autonomous_agent_demo.py --project-dir my-app --max-iterations 5
# YOLO mode: rapid prototyping without browser testing
# YOLO mode: rapid prototyping without testing agents
python autonomous_agent_demo.py --project-dir my-app --yolo
# Configure testing agent ratio (2 testing agents per coding agent)
python autonomous_agent_demo.py --project-dir my-app --testing-ratio 2
# Disable testing agents (similar to YOLO but with verification)
python autonomous_agent_demo.py --project-dir my-app --testing-ratio 0
Authentication:
Uses Claude CLI authentication (run 'claude login' if not logged in)
Authentication is handled by start.bat/start.sh before this runs
@@ -80,7 +90,7 @@ Authentication:
"--max-iterations",
type=int,
default=None,
help="Maximum number of agent iterations (default: unlimited)",
help="Maximum number of agent iterations (default: unlimited, typically 1 for subprocesses)",
)
parser.add_argument(
@@ -94,25 +104,56 @@ Authentication:
"--yolo",
action="store_true",
default=False,
help="Enable YOLO mode: rapid prototyping without browser testing",
help="Enable YOLO mode: skip testing agents for rapid prototyping",
)
# Unified orchestrator mode (replaces --parallel)
parser.add_argument(
"--parallel",
"-p",
"--concurrency", "-c",
type=int,
default=1,
help="Number of concurrent coding agents (default: 1, max: 5)",
)
# Backward compatibility: --parallel is deprecated alias for --concurrency
parser.add_argument(
"--parallel", "-p",
type=int,
nargs="?",
const=3,
default=None,
metavar="N",
help="Enable parallel execution with N concurrent agents (default: 3, max: 5)",
help="DEPRECATED: Use --concurrency instead. Alias for --concurrency.",
)
parser.add_argument(
"--feature-id",
type=int,
default=None,
help="Work on a specific feature ID only (used by parallel orchestrator)",
help="Work on a specific feature ID only (used by orchestrator for coding agents)",
)
# Agent type for subprocess mode
parser.add_argument(
"--agent-type",
choices=["initializer", "coding", "testing"],
default=None,
help="Agent type (used by orchestrator to spawn specialized subprocesses)",
)
# Testing agent configuration
parser.add_argument(
"--testing-ratio",
type=int,
default=1,
help="Testing agents per coding agent (0-3, default: 1). Set to 0 to disable testing agents.",
)
parser.add_argument(
"--count-testing",
action="store_true",
default=False,
help="Count testing agents toward concurrency limit (default: false)",
)
return parser.parse_args()
@@ -120,11 +161,17 @@ Authentication:
def main() -> None:
"""Main entry point."""
print("[ENTRY] autonomous_agent_demo.py starting...", flush=True)
args = parse_args()
# Note: Authentication is handled by start.bat/start.sh before this script runs.
# The Claude SDK auto-detects credentials from ~/.claude/.credentials.json
# Handle deprecated --parallel flag
if args.parallel is not None:
print("WARNING: --parallel is deprecated. Use --concurrency instead.", flush=True)
args.concurrency = args.parallel
# Resolve project directory:
# 1. If absolute path, use as-is
# 2. Otherwise, look up from registry by name
@@ -147,28 +194,35 @@ def main() -> None:
return
try:
if args.parallel is not None:
# Parallel execution mode
from parallel_orchestrator import run_parallel_orchestrator
print(f"Running in parallel mode with {args.parallel} concurrent agents")
asyncio.run(
run_parallel_orchestrator(
project_dir=project_dir,
max_concurrency=args.parallel,
model=args.model,
yolo_mode=args.yolo,
)
)
else:
# Standard single-agent mode (MCP server handles feature database)
if args.agent_type:
# Subprocess mode - spawned by orchestrator for a specific role
asyncio.run(
run_autonomous_agent(
project_dir=project_dir,
model=args.model,
max_iterations=args.max_iterations,
max_iterations=args.max_iterations or 1,
yolo_mode=args.yolo,
feature_id=args.feature_id,
agent_type=args.agent_type,
)
)
else:
# Entry point mode - always use unified orchestrator
from parallel_orchestrator import run_parallel_orchestrator
# Clamp concurrency to valid range (1-5)
concurrency = max(1, min(args.concurrency, 5))
if concurrency != args.concurrency:
print(f"Clamping concurrency to valid range: {concurrency}", flush=True)
asyncio.run(
run_parallel_orchestrator(
project_dir=project_dir,
max_concurrency=concurrency,
model=args.model,
yolo_mode=args.yolo,
testing_agent_ratio=args.testing_ratio,
count_testing_in_concurrency=args.count_testing,
)
)
except KeyboardInterrupt: