# Kaioken Roadmap
A Rust-based HTTP load testing tool with real-time terminal UI and DBZ flavor.
## Vision
Fast local load testing against HTTP endpoints with zero setup friction, real-time visibility, deterministic artifacts, and a memorable DBZ-themed UX. CI/CD ready with checks and thresholds.
## Competitive Positioning
| **Real-time TUI** | ✅ Unique | ❌ | ✅ | ❌ | ❌ |
| **Compare/Regression** | ✅ Unique | ❌ | ❌ | ❌ | ❌ |
| **Zero Config** | ✅ | ❌ | ✅ | ✅ | ❌ |
| **Constant Arrival Rate** | ✅ | ✅ | ❌ | ❌ | ✅ |
| **CI/CD Thresholds** | ✅ | ✅ | ❌ | ❌ | ✅ |
| **Request Chaining** | ✅ | ✅ | ❌ | ❌ | ✅ |
| **Multi-scenario** | ✅ | ✅ | ❌ | ❌ | ✅ |
| **Latency Correction** | ✅ | ❌ | ✅ | ❌ | ❌ |
| **HTTP/3** | ✅* | ❌ | ✅ | ❌ | ❌ |
| **WebSocket** | ✅ | ✅ | ❌ | ❌ | ✅ |
| **gRPC** | ✅* | ✅ | ❌ | ❌ | ✅ |
| **Prometheus Export** | 🔜 v1.4 | ✅ | ❌ | ❌ | ✅ |
| **Language** | Rust | Go | Rust | C | Scala |
### Key Differentiators
**vs k6**: Real-time TUI, regression detection, zero setup, single binary (no runtime)
**vs oha**: Advanced scenarios, request chaining, thresholds, arrival rate, compare mode
**vs wrk**: Lua-free config, multi-scenario, thresholds, rich output formats
**vs Gatling**: Zero JVM overhead, instant startup, real-time feedback
---
## Completed Milestones
### v0.1 — Core ✓
- [x] CLI with clap (url, -c, -d, --timeout, -o, -H, -m, -b)
- [x] Engine: concurrent worker pool, unlimited throughput mode
- [x] HTTP client with reqwest (connection pooling, timeouts, TLS)
- [x] Stats: HDR histogram, status codes, error classification
- [x] TUI: power panel, latency bars, status codes, errors, sparkline
- [x] JSON export with full metrics
- [x] DBZ flavor (`--serious` to disable)
- [x] Headless mode (`--no-tui`, `--json`)
- [x] Safety warning for remote targets
### v0.2 — Load Control ✓
- [x] Rate limiting (`-r, --rate`) with token bucket algorithm
- [x] Ramp-up (`--ramp-up`) - gradually activate workers
- [x] Warmup (`--warmup`) - discard initial metrics, prime connections
- [x] TOML config file support (`-f, --config`)
- [x] CSV output format (`--format csv`)
- [x] Markdown output format (`--format md`)
- [x] Environment variable interpolation in config (`${VAR}`)
### v0.3 — Compare Mode ✓
- [x] `kaioken compare <baseline.json> <current.json>` subcommand
- [x] Side-by-side diff table (RPS, latency percentiles, error rate)
- [x] Regression detection with configurable thresholds
- [x] Exit code 3 on regression (for CI gating)
- [x] Config compatibility warnings
### v0.4 — Advanced Features ✓
- [x] `--max-requests` cap (stop after N requests)
- [x] Body from file (`--body-file`)
- [x] HTTP/2 support toggle (`--http2`)
- [x] Variable interpolation (`${REQUEST_ID}`, `${TIMESTAMP_MS}`)
- [x] DBZ theme cycle in TUI (`t` key) - 6 themes
### v0.5 — Polish & DX ✓
- [x] Weighted scenarios (`[[scenarios]]` in TOML)
- [x] `kaioken init` - generate starter config file
- [x] Shell completions (bash, zsh, fish)
- [x] `--dry-run` mode (validate config without running)
- [x] Man page generation (`kaioken man`)
- [x] HTML report export (`--format html`)
### v0.6 — Checks & Thresholds ✓
- [x] Thresholds in config (`[thresholds]` section)
- [x] Exit code 4 when thresholds fail
- [x] Threshold results in JSON output
- [x] Status checks (`[[checks]]` with `status == 200`, `status in [...]`)
### v0.7 — Load Profiles & Stages ✓
- [x] Stages (`[[stages]]` with duration and target)
- [x] Auto-calculated total duration from stages
- [x] Gradual worker ramp up/down between stages
### v0.8 — Request Chaining & Checks ✓
- [x] Response body checks (`body contains`, `body matches`)
- [x] Response data extraction (`json:`, `regex:`, `body`)
- [x] Variable interpolation with extracted values (`${varname}`)
### v0.9 — Tags, Checks & Sessions ✓
- [x] **Check pass rate metric** - Track % of requests passing checks
- [x] **Check pass rate threshold** - `check_pass_rate = "> 0.95"` for CI/CD
- [x] **Tags** - Label scenarios for filtering in output
- [x] **Cookie jar** - Automatic session handling (`--cookie-jar`)
- [x] **Checks in JSON output** - Per-check and overall pass rates
---
## Upcoming Milestones
### v1.0 — Constant & Ramping Arrival Rate ✓
- [x] **Constant arrival rate executor** - Fixed RPS regardless of response time
```toml
[load]
arrival_rate = 100 # 100 requests/second, VUs scale automatically
max_vus = 500 # Cap on concurrent VUs
duration = "5m"
```
- [x] **Ramping arrival rate** - RPS-based stages with gradual transitions
```toml
[[stages]]
duration = "1m"
target_rate = 100 # Ramp to 100 RPS
[[stages]]
duration = "5m"
target_rate = 500 # Hold at 500 RPS
```
- [x] **Dropped iteration tracking** - Metric when VUs can't keep up with rate
- [x] **Auto VU scaling** - Dynamically spawn/retire VUs to maintain rate
- [x] CLI flags: `--arrival-rate` and `--max-vus`
- [x] TUI display for VUs active/max and dropped iterations
### v1.0.1 — Accuracy ✓
- [x] **Latency correction** - Avoid coordinated omission problem
- Auto-enabled for arrival rate mode, disable with `--no-latency-correction`
- Tracks queue wait time separately from actual server response time
- TUI shows `[corrected]` indicator when active
- [x] **Redirect control** - `follow_redirects = false` option
- [x] **Improved error messages** - Suggestions for common mistakes
### v1.0.2 — WebSocket Support ✓
- [x] **WebSocket connections** - `ws://` and `wss://` protocol support
- [x] **Echo mode** - Send message, wait for response, measure RTT
- [x] **Fire-and-forget mode** - Send-only for throughput testing
- [x] **Connection-based load model** - N connections, each sends at interval
- [x] **WebSocket metrics** - Messages/sec, connection time, latency percentiles
- [x] CLI flags: `--ws-message-interval`, `--ws-fire-and-forget`
- [x] TOML config: `[websocket]` section with `message_interval` and `mode`
### v1.1 — Protocol Expansion ✓
- [x] **HTTP/3 support** - QUIC-based protocol (experimental, feature-gated)
- Enable with `--features http3`
- Uses quinn + h3 for QUIC transport
- [x] **gRPC unary calls** - Request/response pattern
- Enable with `--features grpc`
- Uses tonic for gRPC client
- [x] **gRPC server streaming** - Receive multiple responses
- [x] **CI/CD templates** - GitHub Actions, GitLab CI, Jenkins
- [x] **HAR import** - Convert browser recordings to load tests
- [x] **GitHub release workflow** - Cross-compiled binaries
### v1.4 — Observability & Integration
- [ ] **Prometheus metrics endpoint** - Expose `/metrics` while test runs
- [ ] **InfluxDB export** - Time-series metrics output
- [ ] **OpenTelemetry traces** - Distributed tracing support
- [ ] **Import converters** - Postman collections, OpenAPI specs
### v1.5 — Production Polish
- [ ] **Comprehensive test suite** - Unit, integration, e2e tests
- [ ] **Documentation site** - Dedicated docs with examples and tutorials
- [ ] **Plugin system** - Custom checks, extractors, output formats
- [ ] **Statistical significance** - Multi-run baseline comparison with confidence intervals
---
## Future Considerations (v2.x+)
**Protocol Support:**
- GraphQL — Query-aware load testing with introspection
- MQTT — IoT protocol testing
- Kafka — Message queue load testing
- TCP/UDP — Raw socket testing
**Advanced Features:**
- Distributed mode — Coordinated multi-node load generation
- Lua/Rhai scripting — Dynamic request generation
- File uploads — multipart/form-data support
- Proxy support — HTTP/SOCKS proxy
- Custom metrics — User-defined counters/gauges
**Metrics & Analysis:**
- Keep-alive metrics — Connection reuse tracking
- DNS re-resolution — For DNS-based load balancing
- Flame graphs — CPU profiling integration
- AI-powered anomaly detection — Smart regression detection
- Connection pool metrics — Track reuse, idle, and failed connections
**Developer Experience:**
- VS Code extension — Syntax highlighting for kaioken configs
- Browser extension — Record traffic as kaioken scenarios
- Interactive mode — REPL for exploring responses
---
## Non-Goals (v1.x)
- Browser automation / JavaScript execution
- Distributed coordination requiring external infrastructure
- "Pure server latency" measurement (includes client overhead by design)
- Comprehensive TLS/cert testing matrix
- GUI application (TUI is the interface)
---
## Exit Codes
| 0 | Success |
| 1 | Error (high error rate, config issues) |
| 3 | Regressions detected (compare mode) |
| 4 | Thresholds failed |
| 5 | Load model mismatch in compare (without --force) |
---
## Contributing
Contributions welcome! Please open an issue to discuss significant changes before submitting PRs.
