____ _ _ _ _ ____ ____ ____
/ ___|| \ | | \ | |/ ___|| _ \/ ___|
\___ \| \| | \| | | _ | |_) \___ \
___) | |\ | |\ | |_| || __/ ___) |
|____/|_| \_|_| \_|\____||_| |____/
SYNAPSE deploys 10 specialized AI agents in a sequential pipeline to provide comprehensive code review, security analysis, performance profiling, and automated documentation. Each agent is purpose-built for a single domain, producing deep, actionable findings while consuming a predictable token budget.
Daily throughput: 50–100M+ tokens across thousands of analyses.
- 10 Specialized Agents — Security, quality, performance, architecture, testing, docs, dependencies, refactoring, types, and changelogs
- Deterministic Token Budgets — Per-agent token estimates let you forecast cost before running
- Async Pipeline — Agents execute in sequence; findings from each feed the next
- Web Dashboard — Real-time view of pipeline progress and token consumption
- CLI-First —
synapse analyze src/is all you need
┌─────────────────────────────────┐
│ SYNAPSE CLI │
└───────────────┬─────────────────┘
│
┌────────────────────▼────────────────────┐
│ Pipeline Orchestrator │
└────────────────────┬────────────────────┘
│
┌──────────┬──────────┬────────────┼────────────┬──────────┬──────────┐
▼ ▼ ▼ ▼ ▼ ▼ ▼
┌────────┐ ┌────────┐ ┌────────┐ ┌─────────┐ ┌────────┐ ┌────────┐ ┌────────┐
│Security│ │ Code │ │Perform-│ │Architec-│ │ Test │ │ Doc │ │Depend- │
│Scanner │ │Quality │ │ ance │ │ ture │ │Genera- │ │Genera- │ │ ency │
│ 18K tk │ │ 15K tk │ │ 16K tk │ │ 20K tk │ │ 22K tk │ │ 18K tk │ │ 12K tk │
└────┬───┘ └────┬───┘ └────┬───┘ └────┬────┘ └────┬───┘ └────┬───┘ └────┬───┘
│ │ │ │ │ │ │
▼ ▼ ▼ ▼ ▼ ▼ ▼
┌────────┐ ┌────────┐ ┌────────┐
│Refactor│ │ Type │ │Change- │ ┌──────────────────────────────┐
│Advisor │ │Checker │ │ log │──────▶ │ Aggregated Report │
│ 16K tk │ │ 14K tk │ │ 10K tk │ │ Total: ~161K tokens/file │
└────────┘ └────────┘ └────────┘ └──────────────────────────────┘
| # | Agent | Tokens / Analysis | Capability |
|---|---|---|---|
| 1 | Security Scanner | ~18,000 | OWASP Top 10, injection, XSS, SSRF |
| 2 | Code Quality | ~15,000 | Complexity, maintainability, smells |
| 3 | Performance Analyzer | ~16,000 | Bottlenecks, N+1, memory leaks |
| 4 | Architecture Review | ~20,000 | SOLID, coupling, cohesion, patterns |
| 5 | Test Generator | ~22,000 | Unit tests, integration, edge cases |
| 6 | Doc Generator | ~18,000 | API docs, README, inline comments |
| 7 | Dependency Audit | ~12,000 | CVE scan, license, outdated pkgs |
| 8 | Refactor Advisor | ~16,000 | DRY violations, extract method |
| 9 | Type Checker | ~14,000 | Type inference, null safety |
| 10 | Changelog Generator | ~10,000 | Semver, breaking changes, migration |
Per-file total: ~161,000 tokens · Daily estimate: 50–100M+ tokens
# Clone the repository
git clone https://github.com/nousresearch/synapse.git
cd synapse
# Install in development mode
pip install -e ".[dev]"
# Or install from PyPI
pip install synapse-ai- Python 3.10+
- API key for your preferred LLM provider (OpenAI, Anthropic, or local)
export SYNAPSE_API_KEY="sk-..."
export SYNAPSE_PROVIDER="openai" # or "anthropic", "ollama"# Analyze a single file
synapse analyze src/app.py
# Analyze an entire directory
synapse analyze src/ --recursive --agents security,performance
# List all available agents
synapse agents
# View token consumption statistics
synapse stats
# Launch the web dashboard
synapse dashboard --port 8080import asyncio
from synapse.agents.security_scanner import SecurityScanner
from synapse.agents.code_quality import CodeQuality
async def main():
config = {"provider": "openai", "model": "gpt-4o"}
code = open("src/app.py").read()
scanner = SecurityScanner(config)
result = await scanner.analyze(code, context={"file": "src/app.py"})
print(f"Findings: {len(result['findings'])}")
print(f"Tokens used: {result['tokens_used']}")
asyncio.run(main())from synapse.pipeline import Pipeline
pipeline = Pipeline(agents="all")
report = pipeline.run("src/app.py")
report.save("analysis_report.md")Each agent is a self-contained async module:
class SecurityScanner:
"""Security analysis agent. Token consumption: ~18K tokens per analysis."""
def __init__(self, config):
self.config = config
self.token_estimate = 18_000
async def analyze(self, code: str, context: dict) -> dict:
findings = []
# ... analysis logic ...
return {
"findings": findings,
"metrics": {"risk_score": 8.5},
"tokens_used": self.token_estimate,
}See docs/architecture.md for the full agent specification.
synapse/
├── README.md
├── cli.py # CLI entry point (click)
├── pipeline.py # Pipeline orchestrator
├── agents/
│ ├── __init__.py
│ ├── security_scanner.py # OWASP Top 10, injection, XSS
│ ├── code_quality.py # Complexity, maintainability
│ ├── performance_analyzer.py # Bottlenecks, N+1 queries
│ ├── architecture_review.py # SOLID, design patterns
│ ├── test_generator.py # Unit & integration tests
│ ├── doc_generator.py # API docs, README generation
│ ├── dependency_audit.py # CVE, license compliance
│ ├── refactor_advisor.py # DRY violations, extract method
│ ├── type_checker.py # Type inference, null safety
│ └── changelog_generator.py # Semver, breaking changes
├── examples/
│ └── sample_analysis.py # Example usage script
├── docs/
│ └── architecture.md # Detailed architecture docs
└── tests/
└── ...
# Fork & clone, then:
pip install -e ".[dev]"
pytestPlease read CONTRIBUTING.md before submitting PRs.
Apache 2.0 — see LICENSE for details.
Built by Nous Research · Previous submissions: DeepAudit Engine, DocuForge AI, SentinelAI