Products
Continuous EvalsCompare two skill versions
SkillsPublish and pin team skills
ChallengesScoped from real workflows
Capabilities
Agent evaluationsTest your agent against task cases
Execution tracesInspect available run metadata
Evaluation resultsSee scores, checks, and failures
Compare agent changesMeasure a candidate against a baseline
Versioned agent skillsPin the instructions your team uses
Explore
Worked exampleA release review, case by case
Public challengesBrowse every public challenge
Public skillsBundles to inspect and install
Product directorySearch by product or capability
Contract validationVerify a tool contract locally
Deployment
Versalist CloudAvailable
Your Google CloudPlanned
On-premisesIn development
Local developmentAvailable
Compare options
Works with your harness
Connect your coding agent to Versalist over MCP. Run shell-based harnesses with the CLI.
Claude Code
GitHub Copilot
Codex
Cursor
Antigravity
OpenCode
JetBrains Junie
Bring your own harness
Industries
FinanceFraud, payments and market data
PharmaWorked example: an FDA safety-reporting agent
CyberSecurity and privacy challenges
Frontier scienceResearch and science challenges
Use cases
Catch regressions before an agent change shipsCompare a candidate with what runs today
Turn a workflow into a scored challengeFixed cases and one rubric for every attempt
Test your own agent harness from the terminalAny harness that starts from a shell
Keep one skill revision across repositoriesPublish once, pin in vskill.lock
Turn a failed case into the next skill revisionDraft, test and promote an amendment
Get a release signed offShare or export the review and its decisions
Check a tool before your agent calls itSchema, access and declared effects
All use casesEach one with the page that documents it
How it worksDefine, evaluate, inspect and compare
Approval briefA pilot memo for your budget owner
DocsSetup, CLI, API and MCP reference
IntegrationsCoding agents, model providers and local runs
ToolsWhat a tool can do and access
ChangelogWhat changed and when
ResearchPapers on scoring and evaluation
BlogEssays on environments and evaluation
GuidesTesting and reviewing agent changes
Prompt libraryPrompts for challenge and tool workflows
CommunityBenchmarks and scored episodes
Pricing

Sign in Request a demo

Request a demoRequest demo

Governance and corpus

Evidence from your active workspace.

Loading metrics…

An agent change without a baseline is a guess.

Product updates and new challenges, by email

For developers

Quickstart
CLI reference
API reference
API keys

For coding agents

Agent instructions: /docs/llms.txt
Versalist MCP server: /docs/mcp

Setup

Claude Code
GitHub Copilot
Codex
Cursor
Antigravity
OpenCode
JetBrains Junie

For teams

Request a demo
Pricing
Deployment options
Integrations
Versalist vs Braintrust
Versalist vs LangSmith

Docs
Blog
Research
Guides
Changelog
FAQ
About
GitHub
X
LinkedIn
Privacy
Terms