Changelog

You are looking at a red CI and you want your agent to fix it, but the failure context is far too big to paste into a chat window. A link is 30 characters.

Click Share on any failure in Currents and you get a URL that both people and agents can read.

• Dashboard — Share from the Fix with AI dropdown, for a single test or a whole run
• Slack — Fix with AI on a failing test, Fix all with AI on a failing run
• For people — open the link for a readable failure page, no Currents login needed
• For agents — add .md and any agent fetches the raw context: curl -s https://s.crts.sh/<slug>.md
• One-click handoff into Cursor, Claude Code and Copilot from the share page
• Context is frozen when you click Share, so your agent reads exactly the failure you shared
• Links expire after 24 hours
• Share the same failure twice and you get the same link back

Public by design

A share link is public: no Currents account, which is exactly what lets your agent read one with no credentials. The tradeoff is that anyone holding the URL can read the git information, failure, stack traces and console output included.

So we keep that exposure narrow. Links expire after 24 hours, slugs are 72 bits of randomness, share pages are never indexed or cached, and an expired link returns a blank 404 with nothing to probe. Nothing becomes public on its own either: no page view, no Slack notification and no finished run creates a link, only clicking Share does. The page marks failure content as data rather than instructions, and no handoff action puts your test content into a shell command.

Public sharing is on by default. Admins can turn it off for the whole organization in settings, which refuses new shares and leaves existing links working until they expire.

• Sharing AI context
• AI in Currents
• Currents MCP Server

Slack notifications should tell you what happened without turning into another noisy feed. This update makes run updates easier to scan, keeps threaded messages current, and gives you direct paths from a failure to the next step.

• See run status at a glance — every Slack notification now uses status colors, so failed, passing, cancelled, timed-out, and in progress runs are easier to spot.
• Keep threads short and updated — the first message in a run thread updates as the run progresses. Replies only appear for events that need attention, such as failed tests, flaky tests, cancellations, and timeouts.
• Get accurate final results — multi-group runs now send one final summary per destination, and Slack messages update when late test results change the totals.
• Fix failures from Slack — failure messages include a "Fix with AI" action that opens the existing AI-assisted troubleshooting flow with the relevant run and test context.
• Open the right view in Currents — links from Slack notifications now open Run Summary with the relevant failed, flaky, and group filters already applied.
• Control when success messages arrive — send pass-only notifications to a separate channel, or receive a recovery message when a branch succeeds after a failure.
• Send scheduled reports to Slack — automated reports can now post a compact Slack card with run and test totals, success rate, duration, trends, and links to the filtered Insights views.

References

• Slack App
• Automated Reports

We've redesigned our pricing to better match how teams actually use Currents. Test suites are growing faster than ever, especially with AI-assisted development, and our old pricing structure didn't keep up. Rates are lower across the board, volume discounts are bigger, and we removed most of the friction that made purchasing and scaling harder than it needed to be. User limits, overage penalties, and needing a sales call for enterprise features were all pain points we heard about repeatedly. This update addresses all of them.

• Team plan renamed to Scale with cheaper per-test rates and bigger volume discounts. Self-serve tiers now go up to 500K tests/month (previously 150K).
• New Business plan with SSO, SCIM, and other enterprise features available on self-serve. No sales call needed.
• 50 users included on both Scale and Business (up from 10). Read-only guest users are now free and unlimited.
• Fairer overage billed at the same per-test rate as your plan. You can also upgrade mid-cycle and overage recalculates immediately at the new rate.
• Yearly usage pooling — yearly plans now pool your test volume for the full year instead of resetting monthly. We also increased the yearly discount from 1 month free to 2 months free.

Existing Team customers are being automatically moved to Scale with no action needed.

Read the full announcement: Introducing Currents' New Pricing

Currents is now available right in your editor. The new IDE extension brings CI runs, failure details, and troubleshooting tools into VS Code, Cursor, and compatible forks — no need to open the dashboard.

• CI runs in your sidebar — browse the latest runs, filter by branch, author, tag, or status, and see failures grouped by pull request.
• Fix CI failures without leaving the IDE — the "Fix with agent" action sends the failure to your AI assistant, scoped to a single test, a spec file, or a whole run.
• Rich context for AI agents — each failure ships with the full error message, stack trace, test history, and an error-context snapshot: the page accessibility tree, console output, and network logs captured at the moment of failure. Agents troubleshoot from evidence, not guesses.
• Inline trace files — open the Playwright trace for a failed test in an editor panel to step through actions, inspect DOM snapshots, and read console and network logs manually.
• On-demand context via Currents MCP — the extension automatically registers the Currents MCP server, so agents can fetch additional details about runs, tests, and history as they work — no manual setup.

Install it from the VS Code Marketplace or the Open VSX Registry (Cursor and other VS Code forks), or search for "Currents" in your editor's Extensions panel. You'll need a Currents account and an API key.

References

• Documentation
• VS Code Marketplace
• Open VSX Registry

Share Tests Failure details with your team in Linear directly from Currents. Currents injects all the necessary details about test failures for for humans and agents. No tab switching, no copy-paste.

Enable our native integration with Linear to create or link issues directly from Currents. Once enabled, you can see and edit Linear Issues associated with your tests directly in Currents.

• Full Details for Effective Debugging - Currents injects all the necessary details required to troubleshoot test failures.
• Create or Link Existing Issues - add multiple failures to the same issue, track already affected tests without switching tabs or losing focus.
• Secure, Configurable — based on native secure OAuth integration with Linear, connects once for your entire workspace and configured per project.

When a test fails, the Linear icon is right there. Click it, create a new issue or attach an existing one, and move on. Stack traces, run metadata, and Currents links go with it.

Built to replace your manual Linear workflows:

• Pick a team, project, assign a teammate
• Reference failure details
• Add extra details - done!

References

• Documentation
• Currents App in Linear Directory

We've released @currents/playwright 2.0 with full support for Playwright 1.60 and newer.

We've updated the orchestration workflow to use two commands for better control over test discovery and execution.

• Playwright 1.60+ support: Now fully compatible with the latest Playwright versions
• Improved workflow: Separate test discovery (pwc-p discover) from orchestration execution (pwc-p run)
• Better filtering: Apply Playwright filters (--grep, --last-failed, --project) before orchestration

⚠️ Migration Required

If you're using Playwright 1.60+, you must upgrade to @currents/playwright 2.0. The orchestration workflow has changed to a two-step process. See the migration guide for detailed instructions.

• Orchestration Setup
• CI/CD Examples

You can now add guests to your Currents organization without consuming a billable seat. This makes it easier to bring in stakeholders, reviewers, and cross-functional teammates while keeping paid seats reserved for people who need full write access.

• Invite teammates as Guests even when billable seats are full
• Keep billing and seat limits focused on billable members only

This should make it much easier to collaborate with engineering managers, QA leads, support teams, and other stakeholders who need visibility into test outcomes without taking up a paid seat.

• Manage Team
• Billing & Usage

Organization admins can now set a hard cap on how much extra usage is allowed beyond your plan limit before Currents pauses recording new results.

• Enable Limit Extra Usage in Billing & Usage
• Set a cap as a percentage of your plan limit (e.g. 0%, 50%, 100%)
• CI pipelines and test runners keep running — only recording pauses
• Recording resumes automatically when the usage cycle resets, the cap is raised, or the plan is upgraded
• Available now for all customers on extra usage plans

Enabling extra usage cap

Previously, exceeding your plan limit meant accumulating extra usage fees with no upper bound. With the new cap, you decide the ceiling up front — set it to 0% to stop the moment you hit your plan limit, or to 50% to allow a buffer (for example, up to 15,000 recordings on a 10,000 plan) before recording is paused.

When the cap is reached, your tests continue to run in CI without disruption — Currents simply stops ingesting new results until the next cycle, or until you adjust the cap or upgrade the plan.

This makes budgets predictable while keeping pipelines unaffected.

• Capping Extra Usage documentation
• Billing & Usage

Teams use Currents Actions to conditionally skip, tag, or quarantine tests, so PRs stay green and teams aren't blocked on known flakes or temporary exceptions.

In this release we've made Actions easier to trust and debug by surfacing affected tests in two places.

• On each Action, you can see which tests it hit in your chosen window, open recent executions, and tie them back to the commits they ran on.
• Actions → Affected tests has a project-wide list of tests touched by Actions, with filters and a drill-down so you can see which Actions fired and when—without hopping or guessing.

See which tests each Action touches

• Open any Action and scroll to Affected Tests to see tests that matched in your chosen time range
• Tune the lookback window to focus on recent runs or a longer history
• Expand a test to view recent executions where this Action applied

See all the tests affected by an Action

Open Actions → Affected tests for a project-wide view: every test that matched an Action in your lookback period, with the Actions that applied on each test in one place.

• Use Action status and Action type filters to narrow to active rules, skips, quarantines, tags, or other behaviors you care about
• Search by test title, spec file, or action name to jump to a specific case

This makes it easier for admins and contributors to reason about Actions in context: fewer surprises in CI and faster answers when someone asks, "which tests is this rule actually hitting?"

• Using Currents Actions
• Actions reference

Action Notifications help your change or approach expiration -- before surprises hit your pipeline.

Configure delivery once per project, then get alerts in Slack, email, or both, only for the lifecycle events you care about.

• Slack — Send to a single project channel (uses your existing Slack connection); pick which events trigger a message
• Email — Notify the action creator, org admins, and actions admins, plus optional extra addresses; same event controls as Slack
• Fine-grained events — Choose from creation, disable, re-enable, archive, expiring soon (advance warning before expiration), and expired
• Where to configure — Actions → Settings in the dashboard, alongside the rest of action management (separate from run-level Slack notifications)
• Available now for organizations using Actions

Open Actions → Settings, connect or confirm Slack if you use a channel, turn on Email if you want inbox alerts, then select events. Teams that rely on time-boxed quarantines or expiring skips can lean on expiring soon and expired so ownership and cleanup stay visible without manual polling.

• Documentation