Arcus

by Testsigma

AI-generated test cases, coverage gap detection, and a go/no-go release verdict on every pull request.

About

Arcus connects your GitHub pull requests to a test intelligence layer. When a PR is raised, it scopes test cases to the exact diff, flags coverage gaps, and assembles a test plan ready to execute. Across every sprint, a Release Gate synthesizes test evidence into a definitive Go, No-Go, or Conditional verdict.

What happens when a PR is raised

• AI test generation scoped to the PR diff, not a generic template
• Coverage gap detection highlighting untested areas in the mapped session
• Auto-created test plans pre-populated from PR and Jira metadata
• Re-generation on update keeps test cases current as code evolves
• Release Gate verdict based on Coverage, Pass Rate, and Confidence thresholds you configure

Instant test coverage on every pull request

When a developer raises a PR, Arcus surfaces it immediately in the Unmapped session. Your QA team sees every change, ready to map and generate — no manual triage, no missed PRs:

• AI-generated test cases scoped to the exact diff in each PR, powered by the Generator Agent
• Coverage gap identification highlighting areas of the PR with no existing test coverage in the mapped session
• Real-time sync — test cases appear in Arcus the moment a PR is mapped
• Re-generation on PR update keeps test cases current as code evolves through review

Works inside Claude Code — natively

Arcus ships a Claude Code plugin that hooks into every stage of a coding session. Developers and QA engineers using Claude Code get Arcus coverage automatically, without switching context.

Arcus Claude Code plugin

Install the Arcus plugin in Claude Code and every session — prompts, tool calls, file reads/writes, and subagent boundaries — is captured as structured context and streamed to Arcus for test generation and traceability.

• 🧠 Generator Agent — Creates test cases automatically from Claude Code session context: app behaviour, user workflows, and code changes captured in real time.
• ⚡ Session-based generation — Every Claude Code session maps to an Arcus session. Test cases are generated from what Claude actually did — not from static prompts.
• 🔗 PR + Claude context combined — Raise a PR after a Claude Code session and Arcus merges both signals: the session context and the PR diff — richer test cases, fewer gaps.
• 💬 Slash commands in Claude Code — /arcus:login, /arcus:project, /arcus:map — link sessions to Jira tickets and Arcus projects without leaving the IDE.

The Arcus plugin is available on the Claude Code plugin marketplace. Install it alongside this GitHub App for full coverage across both your IDE workflow and your PR pipeline.

Let Jira do the mapping for you

Include a Jira ticket ID anywhere in your PR title or description, and Arcus maps it to the right sprint session automatically:

• PR with PROJ-123 in the title → auto-mapped to that sprint's session in Arcus
• Test cases generated and gap analysis run without anyone touching Arcus manually
• Works across all standard Jira ticket ID formats (e.g. PROJ-123, TS-456)

No Jira ID? The PR lands in the Unmapped queue, where QA can assign it to any Adhoc or sprint session with one click.

Full session control from Arcus

Arcus gives QA teams a structured home for every PR, without disrupting how developers work on GitHub:

• ⚡ Unmapped queue — Every PR surfaces here first. Nothing is generated until QA intentionally maps it — keeping noise out of your sessions.
• 🎯 Sprint sessions — Jira-linked PRs auto-map here. QA can also manually assign any PR to the right sprint for targeted coverage.
• 📋 Adhoc sessions — For hotfixes, experiments, or PRs without sprint context. Manual assignment from the Unmapped queue in one click.
• 🔍 Coverage gap identification — Once mapped, Arcus highlights exactly which areas of the PR have no test coverage in the current session.

Auto-generate test plans from PR context

Arcus doesn't just generate test cases — it assembles full test plans based on the scope of each mapped PR. Plans are structured, linked to sessions, and ready to execute:

• Scope-aware planning: test plans reflect the actual changes in the PR diff, not a generic template
• Sprint-aligned: plans attach to the mapped sprint session, keeping everything traceable from PR to execution
• Adhoc plans: for PRs without sprint context, Arcus generates a standalone plan under the Adhoc session
• Instant structure: title, description, start/end dates, and linked test runs are pre-populated from PR and Jira metadata
• One-click execution: generated test plans can be run immediately from within Arcus without manual setup

Measure quality with QI metrics

Arcus tracks Quality Intelligence metrics across every session and sprint, giving teams a clear picture of release readiness at any point:

• 📊 Coverage — How much of the sprint's testing surface is approved and release-ready. The testing surface includes both existing approved tests and scenarios identified by the Test Planning Agent as missing. Coverage rises as discovered gaps are addressed by approved tests.
• ✅ Pass Rate — Percentage of executed tests that are currently passing. The most direct quality signal — it tells the team whether the code under test is behaving correctly right now.
• 🔮 Confidence — Of the validation scope the team committed to, how much has been validated trustworthily? Trustworthy means executed AND not flaky. Confidence filters out noise and reflects only solid validation evidence.
• 🚀 Release Readiness — A 0–100 composite score that synthesises Coverage, Pass Rate, and Confidence into a single shippability indicator. It is a gradient, not a verdict — 75, 85, and 95 are meaningfully different. The verdict is the Release Gate's job.
• 🚦 Release Gate — The decisive go/no-go verdict. Where Release Readiness is a continuous score, the Gate is rules-based and definitive. It evaluates configurable thresholds against the upstream metrics and produces one of three states: Go, No-Go, or Conditional.

Connect once, cover everything

Arcus installs in minutes and works across all your repositories:

• Install from GitHub Marketplace and select the repos Arcus should monitor
• Connect your Arcus workspace during setup — no config files, no YAML
• Optionally link Jira in Arcus Settings to activate auto-mapping
• Also install the Arcus Claude Code plugin to extend coverage to your IDE sessions
• Works with public and private repositories

Keep developers in the flow

Arcus works in the background. Developers raise PRs as normal — test cases, gap analysis, test plans, and QI metrics are all generated automatically without interrupting code review:

• No new steps in the developer's PR workflow
• Optional PR status check shows QA mapping and coverage status directly on the pull request
• Jira ticket IDs developers already write become the trigger for auto-mapping and plan generation
• Claude Code users get session-level coverage automatically via the Arcus plugin

Resources

• Documentation: testsigma.com/docs/test-management/