ayulestari722/LogAnalyzerAI

# LogAnalyzerAI 7-Agent Log Analysis and Anomaly Detection orchestration system powered by AI-driven development workflows. ~5900 LOC, ~129 tests. ## Architecture | Agent | Role | LOC | |-------|------|-----| | ParserAgent | Multi-format log parsing (JSON, syslog, Apache, nginx) | 419 | | AnomalyAgent | Statistical anomaly detection (z-score spikes, gaps, rate changes) | 285 | | PatternAgent | Regex pattern matching & frequency analysis | 312 | | CorrelationAgent | Cross-source event correlation by timestamp & session | 298 | | AlertAgent | Severity classification & threshold-based alerting | 276 | | MetricsAgent | Aggregate stats: error rate, latency percentiles, throughput | 264 | | SummaryAgent | Final report aggregation with recommendations | 248 | The orchestrator coordinates the 6 analytic agents via `asyncio.gather()` with a per-agent timeout, then runs `SummaryAgent` once their results are in. Severity-weighted scoring rolls findings up to a single verdict. ## Features - Async parallel agent dispatch with per-agent timeout - Severity scoring: Critical / High / Medium / Low / Info - Multiple output formats: JSON, Markdown, SARIF - Multi-format log parsing (JSON, syslog, nginx access, application logs) - Statistical anomaly detection with z-score analysis - Cross-source event correlation and causal chain detection - Configurable thresholds via YAML - Rich-formatted CLI output - Pluggable agent system via `BaseAgent` inheritance - Watch mode for continuous monitoring ## Usage pip install -r requirements.txt python -m src.cli run ./examples/sample_logs --format markdown python -m src.cli run ./my_logs --format json --output result.json python -m src.cli run ./logs --format sarif --output report.sarif python -m src.cli watch ./logs --interval 5 ## Token Consumption During the design and implementation phase, this project consumed **~12M tokens/day** across Hermes Agent, Claude Code, and Xiaomi MiMo V2.5 Pro for multi-agent reasoning, code generation, refactoring loops, and continuous test maintenance. ## Testing pytest tests/ -v 129 tests covering all 7 agents, orchestrator dispatch, connectors, models, utilities, and CLI integration. ## Project Structure src/ ├── agents/ (7 agent modules + base.py) ├── connectors/ (filesystem, log_parser, stream) ├── models/ (log_entry, analysis_state, schemas) ├── utils/ (config, logger, metrics, retry, serializers, severity) ├── orchestrator.py ├── cli.py └── main.py tests/ (12 test files, 129 tests) config/default.yaml examples/sample_logs/ ## License MIT