# What Viliha is Viliha sits above the AI coding tools your team already uses, such as Claude Code, Codex and Cursor. It does not write the code. It makes sure the code they write follows your company's standard, and it keeps the evidence. ## The problem it solves AI tools write a large share of the code now, in every project, each with its own habits. A company with many projects ends up with as many sets of rules, no shared memory of what was decided, and reviews nobody can replay later. ## What it does - **The record.** Each project keeps its requirements, decisions, rules and status in one place, in git. Every answer cites the file and line it came from. - **The standard.** One set of rules for the whole company, applied to every project, by trigger: security cards for authentication code, release gates for a release. - **The gate.** Inside Claude Code, Viliha checks before every edit that the agent read the rules that apply and stays inside the requirement. - **Review and the Decision Audit.** Every merge request gets a verdict, a score with its doubt, and the evidence behind each factor. A person signs off sensitive changes, and the team's verdicts teach the audit its bar. - **Cited answers.** Anyone on the team asks a question and gets the passages of the record that answer it, with their sources. - **The company model.** Teams, roles, projects and invitations, so an agency runs many client projects from one place. ## Who it is for Agencies and software companies of about 10 to 200 engineers, with several projects under way, already using AI coding tools. Individuals and small teams start free. --- # Quickstart ## 1. Sign up Go to [app.viliha.com](https://app.viliha.com/sign-up), enter your work email and open the sign-in link we send you. No password and no credit card. The first person to sign a company up becomes its administrator. Google, Microsoft and GitHub sign-in are coming next: one click instead of an email. ## 2. Connect your first project During the pilot, we connect your repository's record with you, in a short call. Self-serve setup (connect GitHub or GitLab, pick a repository, indexed in about a minute) is being built. ## 3. Ask Open your project in the sidebar and ask a question. Every answer shows its sources; open one to read the passage it came from. ## 4. Invite your team Settings, Members: invite by email with a role (Administrator, Developer, Product Owner, Tester). Each invitation works once, for seven days, and an administrator can resend or withdraw it. ## 5. Connect your agent See [Agents](/docs/agents/) for Claude Code, Codex and other tools. --- # Viliha with Claude Code, Codex and other agents Without an agent, Viliha is your project's memory. With an agent, it governs the agent's work. ## The loop 1. **Before the task:** the agent asks Viliha for the context: the requirement, the rules that apply, past decisions, all cited. 2. **Before every edit:** the gate checks the agent read the required rules and stays inside the requirement. 3. **While it works:** it asks Viliha instead of guessing a rule. 4. **Before the merge request:** it reviews its own change against the standard. 5. **At the merge request:** the Decision Audit records a verdict and its evidence; a person signs off sensitive changes. 6. **After:** what it learned is proposed for a person to approve. ## Claude Code, today Viliha runs inside Claude Code through its hooks: at session start, on every prompt, before every tool use and at finish. The project's workspace carries a generated `CLAUDE.md` and `.claude/settings.json` that point Claude Code at the record. The tools (context, search, cited answers, proposals) come through the Model Context Protocol (MCP). ## One command for every tool (coming) A hosted MCP endpoint at `https://api.viliha.com/mcp`, with browser sign-in, so nobody copies a token: ``` claude mcp add --transport http viliha https://api.viliha.com/mcp ``` Codex, Cursor, VS Code, and your own agents on the Claude or OpenAI APIs connect to the same address. Each tool's exact setup is published here when it ships and has been checked against that tool's current version. ## How fast Viliha's own work in an answer takes about 25 milliseconds on the server. What you feel is mostly the network, and agents keep one connection open, so tool calls stay well under a second. --- # Pricing | Plan | Price | For | Includes | |---|---|---|---| | Free | $0 | anyone exploring | 1 project, up to 3 members, 100 audits and 1,000 API requests a month; the standard, the gate and the checks; review and the Decision Audit; cited answers; bring your own model key | | Pro | $12 a month | one person | up to 5 projects, unlimited audits, the hosted MCP endpoint, everything in Free | | Team | $24 per user a month | agencies and software companies | unlimited projects and members; teams, roles and invitations; review on GitLab and GitHub; security and DevOps cards; release gates and evidence export; API tokens; single sign-on with Google and Microsoft | | Business | $40 per user a month | larger and regulated companies | everything in Team; SAML and SCIM; audit-profile governance; audit log export; priority support; a choice of data region | | Enterprise | custom | the largest companies | Viliha in your own cloud; custom terms | Yearly billing saves 20%. ## AI agents On every paid plan, agents are billed by the hour of useful work, with no seat fee: $8 junior, $12 mid, $20 senior. ## Models Viliha's checks, audits and cited answers need no AI model. Model use is billed at the provider's price, or you bring your own key. ## Questions **Do I need a credit card for Free?** No. **What happens at a Free limit?** Viliha tells you which limit, and you can upgrade at once. --- # Security and data ## Your code and training **Your code and records are never used to train a model unless your company opts in.** The setting is off by default. Viliha improves from what it owns (its own records and docs, public open-source code) and from anonymous product signals, such as which answers were accepted. ## AI models Cited answers, checks and audits need no model. Nothing from your projects is sent to a model provider until your company turns it on and chooses the provider; then credentials are removed from what is sent, and daily use is capped. ## Where data lives Viliha runs on Amazon Web Services in the United States (us-east-1). A choice of region comes with the Business plan. ## How it is kept apart - Each company's data is isolated in the database itself, by row-level security, not only by the application. - Sessions, sign-in links, invitations and API tokens are stored only as hashes. - Integration credentials are never stored in settings; settings name where the secret is kept. - Every change to members, roles and settings is written to the company's audit log. ## Certifications Viliha is not yet SOC 2 certified. The controls are documented against our security standard, and the audit starts as customers require it. --- # How Viliha compares **Viliha does not replace your AI coding tools. It governs them.** Keep the tools your developers like; Viliha gives the company one standard, one memory and one auditable review across all of them. Comparisons as of November 2026; check each vendor's site for its current offer. | Tool | What it is best at | Where Viliha adds | |---|---|---| | Cursor | an AI editor with agents and review; stronger than Viliha at writing code | one standard and gate across every project and tool, replayable decisions | | GitHub Copilot | the assistant, agents and review inside GitHub | the same rules on GitLab and GitHub, and on work from other tools | | Claude Code | an agentic coding tool in the terminal | Viliha runs inside it: context before the task, the gate before each edit, the audit after | | Devin, Factory | autonomous coding agents | governance of their merge requests, with evidence a person can check | | CodeRabbit | AI code review | review tied to the company's standard and decisions, with a verdict, its evidence and its doubt | | Snyk | security scanning | security rules applied by trigger as the code is written, and the release gates | ## When you do not need Viliha One developer on one project, happy with one tool's rules, gets less from it. Viliha pays off with several projects, several tools and a team that has to show why a change was accepted. --- # Questions ## Is Viliha an AI coding assistant? No. It is the layer above them: it gives your AI tools the context and rules of your project, checks their work, and records the evidence. ## Which tools does it work with? Claude Code today, inside it through hooks and MCP. Codex, Cursor, VS Code and agents on the Claude or OpenAI APIs connect through the hosted MCP endpoint, which is coming. ## Do I need to change how my team works? No new editor. Your requirements and decisions live in a record in git, which Viliha reads; your developers keep their tools. ## Is my code used to train AI models? Not unless your company opts in. The setting is off by default. ## How long does setup take? Signing up takes a minute. During the pilot we connect your first project with you; self-serve setup in about two minutes is being built. ## Where is my data? On AWS in the United States. See [Security and data](/docs/security/).