An SDLC agent for your terminal, and a test engine for your requirements.
SonetX ships as two modules under one product. The SonetX CLI runs a full engineering pipeline from your terminal through a private gateway backed by an AI agent team. SonetX Test Generation turns requirement documents into traceable, self-healing BDD test suites.
Personas, pipeline phases, and traversal depth reflect SonetX's default configuration.
A full AI agent team, one terminal command away.
SonetX CLI runs against your private gateway, not a shared model endpoint. The gateway holds an AI agent team of specialist personas for requirements, architecture, development, QA, code review, security, and docs, plus an orchestrator that plans a run and delegates one stage at a time.
Plan & design
An orchestrator persona plans the run and skips any stage the request does not need: a bug fix can start at implementation, a docs request runs only docs.
The requirements persona captures scope and constraints before any code is written.
Structural decisions and API contracts are drafted before implementation starts.
If a specialist lacks essential information, the stage returns blocked with specific questions instead of guessing.
Build & verify
The developer persona writes the code for the stage the orchestrator assigned.
A dedicated test-engineer persona verifies the change and writes a test report.
A report-only reviewer persona flags issues; its tool allowlist means it can read but never edit code.
A security persona reviews the change before it is allowed to ship.
Ship & track
Documentation is updated as its own stage, not an afterthought.
Every stage's output is written to a durable pipeline folder: requirements, architecture, implementation notes, test and review reports.
A live activity view tracks which persona is active and attributes token usage per persona, even if a run wedges.
Every stage finishes with a status of done, blocked, or needs review, never a silent failure.
Plan a build. Answer a question. Resume where it stopped.
SonetX CLI runs interactively or headless. Ask for a feature end to end and the orchestrator plans the stages; if a specialist is missing information, the run stops and asks, rather than guessing.
Describe a feature and the orchestrator plans requirements through docs, skipping any stage the request does not need.
When a specialist lacks information, the run stops and asks; resume with the answers and it picks up exactly where it left off.
Trigger the same pipeline from CI or a scheduled task with a structured, machine-readable event stream.
Attach MCP servers per persona, so only the specialist that needs a tool, like a QA persona using Playwright, gets it.
From requirement to traceable test suite.
SonetX Test Generation reads requirement documents, specs, and API contracts, builds a knowledge graph of how they relate, and generates BDD test cases that stay traceable back to the requirement that produced them.
Ingest & analyze
Auto-detects and structurally parses OpenAPI, Swagger, Postman collections, and Markdown API docs.
Context overflow recovers via recursive chunking rather than failing the whole document.
Requirements, dependencies, and downstream components are stored as a graph, not just a flat list.
Semantically identical requirements are matched and merged above a similarity threshold.
Reason & generate
Multiple pluggable strategies, chain-of-thought, multi-hop retrieval, document analysis, fit different requirement types.
A multi-phase pipeline plans, discovers scenarios, generates Gherkin, and validates it before handing off.
Authentication, edge cases, non-functional boundaries, and security scenarios are enforced, not left to chance.
A separate critic pass checks generated scenarios before they reach the suite.
Govern & track
Every test case maps back to the requirement, source file, and coverage percentage that produced it.
Requirements without test coverage are surfaced with a recommended scenario, not just a missing checkbox.
When a requirement changes, impact analysis shows which downstream tests and components are affected.
Token usage and cost are tracked per project and per document, not just at the account level.
Governance teams read this part first.
Both modules are built to satisfy the people who own access, spend, and compliance, not just the engineers using them day to day.
- Persona prompts never leave the SonetX CLI gateway; the client only holds a thin, signed stub that cannot be overridden by local instructions.
- Every CLI persona ships with a hard tool allowlist: a report-only reviewer persona cannot edit files or run shell commands.
- Test Generation scopes every query to a project: no requirement, test case, or graph result crosses project boundaries.
- Authentication on both modules is key-based, issued by an admin, with no public free tier and no external sign-in path.
- Token usage and cost are logged per project, per document, and per pipeline run, for audit and chargeback.
What changes when the agent and the test engine are one platform?
Most teams run a coding assistant and a test-writing tool as two separate products. Here is what changes when they share one gateway.
| Area | Without SonetX | With SonetX |
|---|---|---|
| Pipeline execution | Manual handoffs between roles and tools | One orchestrator plans and delegates every SDLC stage |
| Test coverage | Written by hand and drifts from requirements | Generated from requirements and traced back to source |
| Requirement changes | Impact discovered by accident, late | Impact analysis in seconds via the knowledge graph |
| Tooling | A coding assistant and a test tool, unconnected | One gateway, two modules, one audit trail |
Ready to run both modules against your own project?
Tell us whether you are starting with the CLI, test generation, or both, and we will set up a walkthrough on a real pipeline and a real requirement set.
contact@srivasistatech.com