What counts as a verified change, and do products share them?

A verified change is a pull request that merges with a passing quality gate; it bills once no matter how many runs it took — retries, failed changes, and bot commits are free. Products don't share allowances: each product keeps its own 100 per month on Team, and unused ones don't move to another product. Extra changes are totalled across products, so the rate falls as you ship more.

Is Tiden secure and compliant?

Tiden is held by Qase, Inc., which maintains an ISO/IEC 27001 information security management system, SOC 2 Type II reports audited over time, and a public SOC 3 report. Reports are available on request. Higher plans add SAML SSO, SCIM provisioning, and optional self-hosted deployment.

Tiden — Intent verification for coding agents

Tiden is intent verification for coding agents: it maps what your product does by reading your code and writing versioned specs beside it, then checks that agent-written pull requests actually deliver what was asked. Each spec (a "promise") must be proven by a generated test, agent sessions can't close while a promise is open, and specs are redrafted on every merge — even merges that bypass Tiden. Production errors trace back to the release that introduced them, so agents can fix them and add the missing test. Built by Qase, Inc., it works alongside Claude Code, Cursor, Copilot, and Codex, installing via a shell one-liner and connecting to GitHub, Jira, Linear, Notion, and Azure DevOps.

Key Features

Product map with auto-extracted specs

Tiden reads your codebase and writes down each spec, organized by component (e.g., Web App, Public API, Billing, Auth), and marks which specs are proven by a test and which have no test yet.

Test-backed promises that gate agent sessions

Agent sessions start from a working contract, and each intent (e.g., 'Add rate limiting to the public API') is broken into promises that each need a passing generated test — a session can't close while a promise is still open.

Specs that change with the code

Tiden drafts spec updates on every merge, comparing what changed in git against what the code was supposed to do, including merges that skip Tiden entirely.

Branch-level review view

Each branch shows what changed and which tests prove it, with per-branch verdicts like '3 of 3 promises delivered · 0 open'. Reviewers can accept a risk or ask for a test before merge.

Per-release verdicts with error tracing

Every release gets a pass/fail verdict broken down by component, with risk scores. Errors in production point back to the release, and untested issues (e.g., first seen in an earlier release) can be sent to an agent to fix.

Integrations with existing tooling

Connects to GitHub for code, Jira and Linear for issues, Notion, Azure DevOps, and error tracking across environments.

How It Works

Map the product

Tiden reads your code and produces a product snapshot: components, specs, and which promises are proven by tests versus untested.

Verify agent work

Agent sessions start from the spec as a contract; generated tests must pass for each promise before the session can settle and close.

Keep specs in sync

On every merge, Tiden drafts spec updates and surfaces what changed in git versus what the code was supposed to do.

Review and merge

Reviewers see a one-verdict-per-branch summary of what was promised and proven, then merge; the quality gate bills one verified change per merged PR regardless of runs.

Trace production errors

Errors in production point back to the release that shipped them, and agents can be dispatched to fix the issue and add the missing test.

Pros & Cons

Pros

Verifies actual intent (what was asked for), not just green tests — it explicitly targets the gap where agents 'agree with themselves'
Specs are kept versioned beside the code and auto-drafted on every merge, including merges that bypass Tiden
Agent sessions can't close with an open promise, enforcing completion rather than trusting agent self-reporting
Generated tests plus a quality gate give a concrete, billable-per-merge verification signal
Per-release verdicts tie production errors back to the release, creating a fix loop with agents

Who It's For

Best for

Engineering teams shipping agent-written code every day who need proof the agent built what was actually asked
Teams already using Claude Code, Cursor, Copilot, or Codex that want intent verification layered on top of their existing agents
Organizations that need spec/coverage auditing (which behaviors are untested) and release-level accountability for regressions
Teams on GitHub with Jira, Linear, Notion, or Azure DevOps who want specs wired into existing workflows

Not ideal for

Solo developers or very small teams with a single product and light agent usage, where Starter's $49/month may exceed their needs

Use Cases

Engineering teams shipping agent-written code daily who need to verify agents built the right thing, not just that tests pass

Reviewing AI-generated pull requests with an explicit list of promises and proof per branch

Auditing which product behaviors have no test coverage and closing the gaps with generated tests

Tracing production errors back to specific releases so agents can fix regressions and add the missing tests

Keeping living, versioned specs beside the code instead of stale documentation

Pricing

Usage-based per-product pricing with a 14-day free trial on Starter and Team. Starter is $49/month for 1 product (3 repositories, 25 verified changes/month). Team is $199 per product per month (unlimited repositories, 100 verified changes per product/month, then $2.00 per extra change tiering down to $1.00 past 1,000, SAML SSO, impact analysis, 12-month evidence retention). Enterprise is custom, a

Starter

$49 per month

For one product: 1 product, up to 3 repositories, 25 verified changes per month, unlimited developers and agents, intent capture and the product map, generated tests and the quality gate, error tracking traced to the release, evidence kept

Team

$199 per product per month

For several products: everything in Starter plus any number of products with unlimited repositories, 100 verified changes per product per month (then $2.00 per extra change, $1.00 past 1,000), impact analysis, SAML SSO, evidence kept 12 mon

Enterprise

Custom

For many products at scale: everything in Team plus telemetry at any volume, evidence kept to your policy, SCIM provisioning, and self-hosted deployment as a paid add-on. Contact sales.

FAQ

What does Tiden do?

Tiden verifies that coding agents deliver what you actually asked for. It reads your code to build a product map of specs, requires each promise to be proven by a test, drafts spec updates on every merge, and gives one verdict per release — with production errors traced back to the release that introduced them.

Which coding agents does Tiden work with?

Tiden works beside Claude Code, Cursor, Copilot, and Codex. Agent sessions start from the spec as a contract and can't close while a promise is still open.

How is Tiden priced?

Starter is $49/month for 1 product, 3 repositories, and 25 verified changes per month. Team is $199 per product per month with unlimited repositories and 100 verified changes per product per month (extra changes from $2.00 each, down to $1.00 past 1,000). Enterprise is custom. Annual billing saves 20%.

Is there a free trial?

Tiden