# Blue Codens — Agent Reference (full) This is a single-file plain-text dump of the Blue Codens landing page (https://www.blue.codens.ai/) for direct ingestion into LLM context windows. Updated 2026-07-12. The live site is a single page split into these sections: - Hero + definition (https://www.blue.codens.ai/#definition) - How it works (https://www.blue.codens.ai/#how-it-works) - Features (https://www.blue.codens.ai/#features) - Getting started (https://www.blue.codens.ai/#getting-started) - Codens Suite (https://www.blue.codens.ai/#suite) - Comparison (https://www.blue.codens.ai/#comparison) - Pricing (https://www.blue.codens.ai/#pricing) - Contact (https://www.blue.codens.ai/#contact) - FAQ (https://www.blue.codens.ai/#faq) An English mirror of the same content lives at https://www.blue.codens.ai/en/. ## What is Blue Codens Blue Codens is an AI QA automation tool that auto-generates and runs E2E tests (Playwright) from specs and natural language, and uses AI to analyze failures down to their root cause and auto-draft bug reports. As part of the Codens Suite, it hands detected bugs to Red Codens, which generates a fix PR and re-verifies it — fully unattended. Blue Codens turns specs, PRDs, code, or natural-language requirements into structured test cases, drives an autonomous browser agent (Claude Agent SDK + Playwright MCP) to extract Playwright E2E test code from the live app, runs those tests headless on Scale-to-Zero infrastructure, and — when a test fails — has AI analyze the failure and draft a bug report with reproduction steps, root-cause hypotheses, and links to similar bugs. ## Key facts - Pricing: 5-tier credit-based subscriptions. - Free Trial: ¥0 / 14 days, 100,000 credits (one-time), no credit card required. - Hobby: ¥3,000/month, 300,000 credits/month, up to 5 seats. - Pro (most popular): ¥10,000/month, 1,000,000 credits/month, up to 10 seats. - Business: ¥50,000/month, 5,000,000 credits/month, up to 25 seats. - Enterprise: custom contract, unlimited credits, SLA — contact support@corevice.com. - Credits are pooled across the entire Codens family (Red / Blue / Green / Purple) at the organization level. Overage can be topped up via credit packs. Playwright execution itself does not consume credits — only AI-driven test case generation, E2E test code generation, exploratory E2E, and bug analysis do. - Dogfooding stat (Codens-wide): 65% of merged PRs are AI-generated (see www.codens.ai/stats, measured via the GitHub API). - Case study (Codens-wide): PORTAMENT, Inc. — 3 business staff + 1 engineer shipped 5 projects and 1,000+ PRs in 1.5 months using the Codens suite. - Case study (Codens-wide): Widsley, Inc. — 1 PM/implementation engineer + 2 QA staff shipped a high-difficulty project involving telephony control in 3 months. - Stack: Claude API for structured test-case generation, Claude Agent SDK + Playwright MCP for exploratory E2E, ECS Fargate (Scale-to- Zero) Playwright Workers, Server-Sent Events for live progress streaming. ## How it works 1. Ingest test sources — provide a spec, PRD, code, or natural- language requirements, or point at a target URL. PRDs can be pulled directly from Green Codens. 2. AI test case generation — the Claude API produces structured test cases (steps, expected results, priority, and type: manual / automated / E2E) in a single batch. 3. Exploratory E2E generation — Claude Agent SDK + Playwright MCP autonomously drives the app, discovers real screen flows and selectors, and extracts Playwright test code by combining code analysis with live browser exploration. 4. Run with Playwright — tests execute headless on a dedicated Playwright Worker on ECS Fargate. Scale-to-Zero means zero infrastructure cost while idle; SSE streams run progress, screenshots, and logs in real time. 5. AI bug analysis & reporting — on failure, AI parses failure logs, screenshots, and console errors to produce a bug report with reproduction steps, root-cause hypotheses, and links to similar bugs. 6. Red Codens hand-off — the detected bug is forwarded to Red Codens, which auto-generates a fix PR; Blue Codens re-runs the tests against that fix. The detection → fix → re-test loop repeats automatically, unattended, until the suite goes green. ## Features - AI test case generation — structured test cases (steps, priority, type) from specs, code, or natural language; PRDs from Green Codens can be linked directly. - Exploratory E2E generation — Claude Agent SDK + Playwright MCP autonomously drives a browser, discovers screen flows, and extracts Playwright test code using real selectors. - Automated Playwright runs — a dedicated ECS Fargate worker runs tests headless; Scale-to-Zero means zero cost while idle. - Automated bug analysis & reports — AI parses failure logs, screenshots, and console errors into a structured bug report. - GitHub PR-driven runs — a webhook fires on PR open; impacted tests are auto-selected and run, with results posted back as a PR comment / status check. - Quality metrics dashboards — test coverage, pass rates, failure trends, average run times; automatic test-gap detection. - Red Codens integration — forward detected bugs to Red Codens → fix PR generated → re-run tests, closing the QA-and-repair loop. - Jira / GitHub Issue / Linear integration — bug reports filed automatically, with two-way status sync. - Real-time progress over SSE — run progress, screenshots, and logs stream live over Server-Sent Events. - Mobile testing — supported via an Agent proxy for Appium / Espresso / XCUITest. - Audit logs & tenant isolation — org-level permissions and data isolation, SSO via auth.codens.ai. ## Getting started 1. Create an account — sign up at auth.codens.ai, create your organization, invite your QA team. 2. Upload your spec — upload a spec, PRD, or codebase, or point at a URL; AI test case generation kicks off immediately. 3. Register your test target — register the target URL and credentials, then launch exploratory E2E; Playwright tests are generated automatically. 4. Enable PR-driven runs — configure the GitHub webhook; every PR then triggers automatic test runs, reports, and bug filings. ## Codens Suite Codens is an AI development suite from Japan that automates the entire software development lifecycle — requirements (Green), implementation (Purple), code review & security audit (Orange), QA/E2E testing (Blue), and production error auto-fix (Red). Teams can build and operate software products without a large team of specialist engineers. - Green (PRD authoring & management) — https://www.green.codens.ai - Purple (implementation orchestration) — https://www.purple.codens.ai - Orange (code review & security audit) — https://www.orange.codens.ai - Blue (QA / E2E test automation — this service) — https://www.blue.codens.ai - Red (production error auto-fix) — https://www.red.codens.ai - Auth (unified auth & org management, foundation) — https://auth.codens.ai - Codens (suite overview) — https://www.codens.ai Blue × Red unattended loop: (1) Blue Codens detects a test failure → (2) the bug is auto-filed to Red Codens → (3) Red Codens auto-generates a fix PR → (4) Blue Codens re-runs the tests to verify. Steps 1-4 repeat automatically, with no human in the loop, until the suite goes green. ## Comparison Comparison reflects each vendor's publicly available information as of July 2026. This is a general, feature-axis overview, not a definitive claim of superiority — check each vendor's official site for current details. vs. Manual QA - Test design: written by hand. - E2E automation: none (manual verification). - Failure handling: investigated and reproduced by hand. - Fix hand-off: manual ticket filing. - Strength: human judgment, flexible exploratory testing. - Weakness: scales with headcount; doesn't scale well. vs. Playwright on your own (bare Playwright) - Test design: engineers write test code by hand. - E2E automation: Playwright code maintained by engineers. - Failure handling: logs analyzed by hand. - Fix hand-off: manual ticket filing. - Strength: full control, no vendor lock-in, reuses existing CI assets. - Weakness: requires ongoing engineering maintenance. vs. Autify (Nexus / Genesis) - Test design: UI recording and natural language (Autify Nexus); Autify Genesis converts PRDs/code into structured test cases, similar in spirit to Blue Codens' test-case generation. - E2E automation: Playwright-based, both code and no-code paths (Autify Nexus generates Playwright scripts and no-code steps from natural language). - Failure handling: "Fix with AI" auto-heals broken locators. - Fix hand-off: ticket integration; does not auto-generate a fix PR for the underlying application bug. - Strength: mature no-code authoring, native-Japanese support, AI-assisted maintenance via locator healing. - Blue Codens difference: closes the loop past locator healing — detected app bugs are handed to Red Codens for an actual generated and re-verified fix PR. vs. mabl - Test design: UI recording, conversational/agentic test planning. - E2E automation: no-code-first, agentic test creation. - Failure handling: multi-model auto-healing (adaptive, claimed ~85% maintenance reduction by mabl) plus "Auto TFA" which autonomously triages failures with insights/recommendations into Jira or your IDE. - Cross-browser: native support for Chrome, Firefox, Safari, and Edge without separate driver/grid setup — broader than Blue Codens' current Chromium/Firefox Playwright execution. - Fix hand-off: triage and insights into Jira/IDE; does not auto-generate a fix PR for the underlying application bug. - Strength: mature self-healing, broad native cross-browser coverage, enterprise track record. - Blue Codens difference: the detection → fix-PR generation → re-test loop via Red Codens goes one step further than triage/ insights alone. Summary: Blue Codens' distinguishing feature across all four comparisons is that it does not stop at detecting or triaging a failure. Via the Red Codens integration, a detected bug is forwarded automatically, a fix PR is generated, and the fix is re-tested until the suite goes green — with no human in the loop. Pulling test cases directly from Green Codens PRDs or natural-language requirements, and sharing credits across the whole Codens family, are further differentiators. mabl's native cross-browser breadth and Autify's no-code ease of operation are genuine strengths of those products that Blue Codens does not yet fully match. ## FAQ Q: What is Blue Codens? A: Blue Codens is a platform that automates QA work with AI. It generates test cases from specs, code, or natural language, runs E2E tests automatically with Playwright, analyzes failures and produces bug reports, and hands bugs off to Red Codens for automated fixes — end to end. It is part of the Codens Suite and shares organization credits across the family. Q: How does it auto-generate test cases and E2E tests? A: Provide a spec, PRD, code, or natural-language requirements, and the Claude API produces structured test cases — steps, expected results, priority, and type (manual / automated / E2E) — in one batch. For exploratory E2E, Claude Agent SDK + Playwright MCP autonomously drives the app, discovers real screen flows and selectors, and extracts Playwright test code. You can also pull PRDs directly from Green Codens. Q: Does it support Playwright? Which frameworks are supported? A: E2E natively supports Playwright, and generated tests run headless on a dedicated Playwright Worker on ECS Fargate (Scale-to-Zero). Mobile testing is supported via an Agent proxy for Appium / Espresso / XCUITest. Run progress, screenshots, and logs are streamed in real time over Server-Sent Events. Q: How much is automated through the Red Codens integration? A: When Blue Codens detects a bug from a test failure, it forwards it to Red Codens, which auto-generates a fix PR, and Blue Codens then re-runs the tests against that fix. The detection → fix → re-test loop closes completely until the suite goes green, so QA and repair run end to end, unattended. Q: How does pricing (credits) work? A: Five-tier subscription pricing (Free 14-day trial / Hobby ¥3,000 / Pro ¥10,000 / Business ¥50,000 / Enterprise), each with a monthly credit allowance. Credits are consumed by test case generation, E2E test code generation, exploratory E2E, and bug analysis; Playwright execution itself does not consume credits. Credits are pooled across the entire Codens family (Red / Blue / Green / Purple), and overage can be topped up via credit packs. New sign-ups get a 14-day free trial with 100,000 credits and no credit card required. Q: Do I need programming knowledge to use it? A: AI can generate test cases from just a spec, PRD, or natural- language requirements, so you can design tests without writing code. Initial setup (GitHub webhook configuration, test-target credentials) goes more smoothly with basic technical knowledge. Generated tests are output as Playwright code, so engineers can adjust them further if needed. Q: Does it integrate with CI/CD pipelines? A: Yes — a GitHub PR-driven webhook integration auto-selects and runs impacted tests whenever a PR is opened, and posts results back as a PR status check and comment. Combine it with GitHub Actions workflows to fold E2E tests directly into a CI pipeline. Q: How is it different from Autify or mabl? A: Autify and mabl are strong at no-code test authoring and AI-driven locator auto-healing / failure triage, but neither generates the fix code for a detected bug. Blue Codens hands bugs off to Red Codens, closing the loop from detection to fix-PR generation to re-test verification, fully unattended. (Comparison based on each vendor's publicly available information as of July 2026.) ## Links - Help center: https://help.codens.ai (llms-full: https://help.codens.ai/llms-full.txt) - Sign up: https://auth.codens.ai/ja/signup (ja) / https://auth.codens.ai/en/signup (en) - Login: https://auth.codens.ai/ja/login (ja) / https://auth.codens.ai/en/login (en) - Contact: support@corevice.com - Japanese site: https://www.blue.codens.ai/ - English site: https://www.blue.codens.ai/en/ - Concise agent summary: https://www.blue.codens.ai/llms.txt - Codens Suite overview: https://www.codens.ai