millionco/expect
TypeScript No license Dev Tools License not recognized by GitHub no-tests-mentionedExpect tests your agent's code in a real browser
3.5k
Stars
153
Forks
22
Open issues
9
Contributors
AI Analysis
Expect is a specialized testing tool that integrates with AI coding agents (Claude Code, Codex, Cursor) to automatically test web application changes in a real browser using Playwright. It reads git diffs, generates test plans, and runs them with real user simulation to catch performance, security, design, and functionality issues. This tool is specifically designed for developers using AI-assisted coding workflows who need automated QA validation without writing test scripts.
Inferred from signals mentioned in the README (tests, CI, type safety) — not a review of the actual code.
AI's overall editorial judgment — not an average of the bars above, can weigh other factors too.
Agent-native browser testing tool that auto-generates and runs tests from code changes
Expect is a testing skill designed to integrate with LLM-based coding agents (Claude Code, Codex, Cursor, etc.). It reads git changes, generates a test plan, and executes it in a real browser using Playwright, checking for performance, security, design, and completeness issues. Adoption appears concentrated among developers using agent-assisted coding workflows. Created March 2026, it remains very early-stage with modest but growing engagement.
Expect launched in March 2026 as a specialized skill for agentic coding workflows. It emerged in response to the rapid adoption of AI coding assistants and the need for automated verification of agent-generated changes without manual test script maintenance.
The project gained 3,523 stars in approximately 3.5 months with a recent pace of 15 stars/week. Growth trajectory suggests early-adopter traction within the agent-assisted development niche, though absolute numbers remain modest. No evidence of viral adoption or mainstream breakthrough yet.
Adoption not verified. No documentation of production deployments, case studies, user testimonials, or community benchmarks. NPM download statistics referenced in README via badge but actual numbers not provided. Demo site (expect.dev) mentioned but cannot verify traffic or usage. Adoption appears limited to early adopters experimenting with agent-assisted workflows.
Based on README, Expect likely wraps Playwright automation and integrates via CLI hooks into multiple agent platforms (Claude Code, Copilot, Cursor, Gemini, Codex, OpenCode). It appears to parse git diffs, generate test plans via LLM, extract browser cookies, and report results back to the agent. The --ci flag suggests headless execution and GitHub Actions support.
Not documented in README. No mention of internal test suite, coverage metrics, or test strategy.
Last push 2026-05-06 (56 days before analysis date 2026-07-01), indicating active but not highly frequent updates. Repository is 3.5 months old; early-stage projects often show episodic rather than continuous commits. No public issue tracker visibility provided. Package is published to npm (expect-cli). Maintenance status is defensible for a new project but too early to assess long-term commitment.
ADOPT IF: you are actively using Claude Code, Copilot, Cursor, or similar agents for web development and want automated regression testing without maintaining selectors or test scripts — and you are comfortable with tooling that is 3.5 months old with unverified production usage. AVOID IF: you need stable, battle-tested testing infrastructure, require extensive integrations beyond listed agents, need mobile testing (not yet available), or work in teams using traditional CI/CD pipelines not tied to agent workflows. MONITOR IF: you are evaluating agentic development workflows long-term; this project may consolidate or diverge significantly as the agent-assisted development category matures, and enterprise/hosted offerings are explicitly pending.
Independent dimensions
Mainstream potential
4/10
Technical importance
6/10
Adoption evidence
2/10
- Very early-stage project (3.5 months old) with no documented production deployments; breaking changes or abandonment possible.
- Adoption not verified beyond implied early adopter base; success depends on agent assistant adoption rates, which are still uncertain.
- Tight coupling to specific agent platforms (Claude Code, Copilot, Cursor, etc.) may limit portability if agent ecosystem consolidates or shifts.
- Missing features (mobile testing, enterprise hosting, formal SLA) may drive users toward alternatives as requirements mature.
- Test generation quality depends on underlying LLM; unclear how Expect handles hallucinations, false positives in test plans, or edge cases from README alone.
Expect will likely remain a specialized skill within agentic development workflows, gaining modest adoption among early adopters of AI-assisted coding. Its trajectory depends heavily on whether AI coding agents become standard developer tools (plausible but not yet inevitable by mid-2026). Enterprise/hosted version and mobile testing are probable next milestones. Project is unlikely to replace Playwright or Cypress but may carve a sustainable niche if agent adoption accelerates.
Newsletter
Get analyses like this every Monday
Free weekly digest of the most interesting open-source discoveries.
Languages
Information
- Website
- https://expect.dev
- Language
- TypeScript
- License
- NOASSERTION
- Last updated
- 2mo ago
- Created
- 4mo ago
- Analyzed with
- anthropic/claude-haiku-4-5
Stars over time
Contributors over time
Top 100 contributors only — repos with more will plateau at 100.
Open issues
browser-mcp.js leaks orphan processes when parent agent exits abnormally — root cause + fix
MCP browser daemon launches stale/ghost Chrome sessions, requires manual pkill to recover (~50% failure rate)
expect-cli crashes on launch: ERR_UNSUPPORTED_DIR_IMPORT (ink → es-toolkit/compat)
expect keeps trying to open dia (instead of chrome)
fix: rename README.md to SKILL.md when copying skill directory
Top contributors
Similar repos
vercel-labs/agent-browser
agent-browser is a fast native Rust CLI for browser automation designed...
ColeMurray/background-agents
Background Agents is an open-source hosted system for running AI-powered coding...
| Repository | Stars | Week Δ | Language | Score | Updated |
|---|---|---|---|---|---|
|
millionco/expect
this
|
3.5k | +3 | TypeScript | 7/10 | 2mo ago |
 microsoft/playwright
|
92.6k | — | TypeScript | 10/10 | 5h ago |
|
vercel-labs/agent-browser
|
38.2k | — | Rust | 7/10 | 2d ago |
 bug0inc/passmark
|
1.1k | — | TypeScript | 7/10 | 1d ago |
|
ColeMurray/background-agents
|
2.1k | — | TypeScript | 7/10 | 8h ago |
 hyperbrowserai/HyperAgent
|
1.5k | — | TypeScript | 7/10 | 2mo ago |
Mature, language-agnostic browser automation framework with 92k stars. Expect builds on Playwright but adds LLM-driven test generation and agent integration; Playwright requires manual test script authoring.
4,225 stars, established BDD testing framework. More mature and broader adoption, but requires developer-written scenarios; Expect aims to auto-generate tests from code changes.
5,563 stars, also bridges Playwright and agentic workflows via MCP. Expect appears more opinionated toward test generation; comparison of architectural approach is unclear from README alone.
1,447 stars, general-purpose browser agent. Expect is narrowly scoped to testing and integrates deeply with multiple coding agents; HyperAgent appears more general-purpose.
Not listed but implicit competitor. Cypress is mature and widely adopted for E2E testing; Expect differentiates by eliminating manual selector maintenance and test script ownership.
