Sverka — User Documentation

Sverka is a framework for running local checks. Define checks in TypeScript, run them locally, get structured findings. Compile to CI when you need to.

ConceptsCopy Anchor Link

• Why Sverka — what it solves, how it fits, design principles
• Use cases — local checks, CI compilation, agent orchestration, SARIF tooling
• Playground — browser pipeline runner, share links, docs embeds
• Benchmark — two-agent comparison harness + static dashboard

Getting startedCopy Anchor Link

• Install — prerequisites, bun install, sverka init
• First workflow — define, run, view, compile

Running checksCopy Anchor Link

• CLI — all commands, global flags, exit codes
• Saga compensations — automatic rollback of succeeded steps on failure
• Suspend and resume — pause runs for external input, resume with data
• Run queries — read-only snapshot of run state
• Snapshot storage — persistent storage for suspend/resume snapshots

Findings & SARIFCopy Anchor Link

• SARIF pipeline — serialize findings, view in TUI, generate HTML, web dashboard

Remote hubCopy Anchor Link

• Remote run hub — self-hosted shared cache + run history, tokens, /v1/ API, dashboard
• Embedding the hub — @sverka/hub programmatic API, store, dashboard pages

WorkflowsCopy Anchor Link

• Overview — Construct API authoring surface

Agent integrationCopy Anchor Link

• Skill + CLI — AI agent integration via skill and CLI
• MCP server — Sverka as an MCP server and MCP plugin client

Compiling to CICopy Anchor Link

• GitHub Actions — compile to GitHub Actions YAML
• GitLab CI — compile to GitLab CI YAML
• Agentic workflows for GitLab — gh-aw-style agent jobs, comment/issue triggers, safe-outputs
• GitLab webhook setup — webhook → pipeline-trigger contract, schedules, apply token

Hosted executionCopy Anchor Link

• Bootstrap job — run the whole pipeline inside one CI job with sverka run

ReferenceCopy Anchor Link

• Verification package — @sverka/verification: checks + findings + policy in one package
• Built-in checks — check IDs, resolver behavior, SARIF extraction
• Findings normalization — SARIF normalization, fingerprints, baselines
• Policy enforcement — rules, severities, enforcement
• Run reporting — @sverka/reporter: text/HTML/TUI renderers, reducer, policy gate
• Run audit (planned) — per-step timings, AI cost estimation
• Graph visualization (planned) — Mermaid flowchart output
• Markdown authoring — .sverka.md files
• Roadmap — planned features and future targets