Mvp v1

[n2400813g]

Summary by cubic

Launches blop v0.2.0 (MVP): an MCP-based QA tool that discovers, records, and replays web app flows with SQLite-backed runs and optional HTTP streaming. Includes a full tool surface, resilient replay with auto-heal, and CI with tests.

• New Features

  • MCP server (blop) exposing tools for discovery, recording, regression, results, validation, auth, and a v2 resource surface.
  • Resilient engines: BFS discovery, guided recording with assertions/screenshots, hybrid regression with agent fallback, and severity classification.
  • Pluggable LLMs via BLOP_LLM_PROVIDER (google default; anthropic/openai via extras). Dev helper blop-inspect for the MCP Inspector.
  • Storage: SQLite (.blop/runs.db) plus artifact dirs under runs/; optional HTTP SSE server via blop[server] (blop-http).
  • Auth: env-login, storage_state, and cookie_json profiles, plus interactive capture_auth_session.
  • Tooling: .env.example, .cursor/mcp.json.example, GitHub Actions CI (pytest, Playwright, ruff), and MIT LICENSE.

• Migration

  • Package renamed from vibetest to blop; a shim keeps vibetest entry points, but new imports should use blop.*.
  • Install with uv pip install -e .; install Playwright browsers; set GOOGLE_API_KEY or BLOP_LLM_PROVIDER with the matching key, then copy .env.example → .env.
  • Paths: artifacts under runs/ and DB at .blop/runs.db (override with BLOP_RUNS_DIR/BLOP_DB_PATH).
  • For streaming, install blop[server] and run blop-http; for interactive tool testing, run blop-inspect.

^{Written for commit 92b5932. Summary will update on new commits.}

[cubic-dev-ai]

No issues found across 87 files

Note: This PR contains a large number of files. cubic only reviews up to 75 files per PR, so some files may not have been reviewed.