CovSight is an open-source coverage intelligence platform built on open data formats. It ingests coverage from any simulator — VCS, Questa, Xcelium, Verilator, cocotb, and more — stores it in an open, portable format, links it to a structured testplan, and delivers actionable insights through the terminal, VS Code, CI pipelines, and AI agents.
Your coverage data is yours: open formats, open APIs, no walled gardens.
Commercial EDA tools lock coverage data in closed, simulator-specific formats. Open-source EDA has no equivalent at all. Teams mixing Verilator for CI with a commercial simulator for signoff have no unified data model. CovSight breaks down those walls:
| Without CovSight | With CovSight |
|---|---|
| Coverage locked in closed, simulator-specific formats | Open formats — read and write from any tool |
| Analysis requires a proprietary vendor GUI | Terminal, IDE, or CI — your choice |
| Testplans live in spreadsheets or vendor GUIs | Version-controlled YAML in your repo |
| Regressions produce GBs of raw vendor data | Compact, portable NCDB artifacts — KBs not GBs |
| Coverage history locked in vendor tools; trends lost | Persistent open test history with trend analysis |
CovSight reads coverage from VCS, Questa, Xcelium, Verilator, cocotb, and any UCIS-compliant source. It writes to LCOV, Cobertura, JaCoCo, and Clover for software CI integration. Conversion is lossless where formats allow, with explicit warnings when fidelity is reduced.
NCDB — CovSight's open binary format — achieves 60–73× size reduction over SQLite through LEB128 variable-length integers, toggle-pair compression, and schema-aware encoding. A 150 MB UCIS XML database becomes under 2 MB. NCDB files are plain ZIP archives: portable, diffable, and safe to commit to version control or store as CI artifacts.
CovSight defines a YAML-based superset testplan format that imports from OpenTitan Hjson, Cadence VPF XML, Synopsys VC Planner, and Siemens Questa Visualizer formats. Testplans live in the repo, are version-controlled, support wildcard expansion, composable imports, and direct bindings to coverage database paths for automated closure computation.
CovSight's analysis engine tracks per-test coverage contributions over time, identifies flaky tests using statistical scoring, and detects regressions with CUSUM change detection. The MCP server surfaces this data to AI agents, enabling natural-language queries about coverage gaps, test contributions, and what to verify next.
| Interface | Description |
|---|---|
| CLI | covsight convert, merge, report, testplan — scriptable, composable, CI-ready |
| TUI | Interactive terminal UI with dashboard, hierarchy tree, coverage gaps, hotspots, and metrics views |
| CI Pipelines | Use the CLI in any CI environment — GitHub Actions, GitLab CI, Jenkins |
| AI Agents (MCP) | MCP tools expose coverage databases to LLM agents for natural-language analysis |
| Python API | Full programmatic access to all coverage data, analysis, and format conversion |
| VS Code Extension | Hierarchy tree, source gutter decorations, and a dashboard webview (coming soon) |
| Package | Description | Install |
|---|---|---|
| covsight | CLI · TUI · MCP server · format adapters | pip install covsight |
| covsight-core | UCIS data model · NCDB format · merge engine · testplan | pip install covsight-core / npm install @covsight/core |
pip install covsightcovsight includes the CLI, TUI, MCP server, and all format adapters. covsight-core is the underlying library for programmatic use.