v0.0.6-alpha

Bubblegum v0.0.6-alpha — first PyPI release + JSON-RPC engine bridge

First Bubblegum pre-release published to PyPI, plus the foundation for using
the engine from non-Python projects.

Highlights

• 📦 On PyPI: pip install bubblegum-ai (0.0.6a0).
• 🌉 Engine bridge (bubblegum.bridge) — a JSON-RPC 2.0 server (bubblegum bridge / python -m bubblegum.bridge) that exposes the four primitives
  (act / verify / extract / recover) over stdio, with version/capability
  negotiation. One Python engine; other languages drive it without re-implementing
  grounding.
• 📘 Docs: focused Web and Mobile how-to guides + a dual-distribution
  (PyPI + npm) strategy.
• 🔁 Self-healing, memory cache, and the deterministic-first resolver chain,
  unchanged and stable.

Companion: TypeScript/JS client

A Node client, @bubblegum-ai/node, is published on npm so JS/TS Playwright
suites can drive the same engine — npm install @bubblegum-ai/node. See
docs/HOW_TO_USE_TYPESCRIPT.md. (Client + CDP-attach support landed in commits
after this tag.)

Install

pip install "bubblegum-ai[web]"   # + Playwright
# or, from TS/JS:
npm install @bubblegum-ai/node

Bubblegum v0.0.5-alpha

v0.0.5-alpha

• Release scope finalized for GitHub pre-release v0.0.5-alpha with package version 0.0.5a0 (PEP 440).
• Scope includes:
  • Phase 17A roadmap reset and v0.0.5-alpha planning
  • Phase 17B real smoke kit/adoption readiness audit
  • Phase 17C real smoke kit docs/examples MVP
  • Phase 17D smoke runner audit
  • Phase 17E dependency-free infra-free smoke runner MVP
  • Phase 17F smoke runner post-merge verification
  • Phase 17G release checklist collect-only baseline sync to 615
  • Phase 18B release metadata/docs/checklist preparation
• No runtime behavior changes.
• No SDK public API changes.
• No schema changes.
• No dependency changes.
• No provider/network/browser/device CI smoke added.
• PyPI/TestPyPI publishing remains deferred; release target remains GitHub pre-release only.

Bubblegum v0.0.4-alpha

v0.0.4-alpha

1. Release scope finalized for GitHub pre-release v0.0.4-alpha with package version 0.0.4a0 (PEP 440).

2. Scope includes:

   • Phase 14 adoption docs/examples polish
   • Phase 15B adapter-level transient retry MVP
   • Phase 15D retry observability metadata/reporting
   • Phase 15F adapter-level explicit wait_for MVP
   • Phase 15H wait observability metadata/reporting

3. No SDK public API changes.

4. No schema changes.

5. No dependency changes.

6. No provider/LLM/OCR/vision retry behavior changes.

7. PyPI/TestPyPI publishing remains deferred; release target remains GitHub pre-release only.

Bubblegum v0.0.3-alpha

Bubblegum v0.0.3-alpha

Highlights

This pre-release focuses on visual reference hydration and reporting readiness.

Added

• VisualRefHydrator abstraction for safe synthetic visual reference handling.
• Deterministic web hydration for OCR and vision metadata.
• Deterministic mobile hydration using Appium hierarchy XML with text, content-desc, and resource-id matching.
• Sanitized hydration diagnostics on SDK result metadata.
• JSON and HTML report visibility for hydration diagnostics.
• Hydration analytics summary in report analytics.
• Release checklist and package metadata aligned for package version 0.0.3a0.

Safety and privacy

• Synthetic ocr:// and vision:// refs are not executed directly.
• Hydration performs no screenshot/provider calls.
• No bbox proximity or center-tap fallback is enabled.
• Reports redact unsafe fields such as hierarchy XML, screenshots, image bytes, base64, raw payloads, provider bodies, secrets, and candidate dumps.

Validation

• Package version: 0.0.3a0
• Static benchmark: 12/12
• Execution benchmark: 12/12
• Collect-only baseline: 584 tests
• Clean build and twine check expected before release.

Publishing

• GitHub pre-release only.
• PyPI/TestPyPI publishing remains deferred.

Bubblegum v0.0.2-alpha

Bubblegum v0.0.2-alpha

Highlights

• Added callable-only OCR backend support with opt-in privacy gating.
• Added VisionProvider abstraction and callable vision provider support.
• Added optional OpenAIVisionProvider backend without making OpenAI a mandatory dependency.
• Added public vision provider lifecycle APIs: configure_vision_provider(...) and clear_vision_provider().
• Added SDK-level screenshot-to-vision wiring with strict privacy and cost gates.
• Added sanitized OpenAI vision diagnostics.
• Added manual OpenAI vision provider usage example.

Vision and OCR posture

• OCR remains callable-only; no bundled real OCR dependency is included.
• Vision provider usage is optional and dependency-light.
• Provider-based screenshot vision requires:
  • grounding.enable_vision=true
  • privacy.send_screenshots=true
  • privacy.process_screenshots_for_vision=true
  • max_cost_level="high"
  • registered provider
  • screenshot bytes present
• Manual vision_candidates remain supported and are not cost-blocked.
• vision://... refs remain synthetic and non-executable.

Safety and privacy

• Raw screenshots, base64 payloads, API keys, and secrets are not persisted or exposed.
• OpenAI provider diagnostics are sanitized.
• Provider failures fail safe to empty candidates.

Validation

• Package validation passed.
• Strict validation passed.
• Build passed.
• Static benchmark: 12/12.
• Execution benchmark: 12/12.
• Targeted Phase 11J/11N/11R/11X tests passed.
• Public API, packaging extras, and package metadata tests passed.
• pytest --collect-only -q: 539 tests collected.

Known limitations

• PyPI/TestPyPI publishing is deferred.
• Playwright/Appium runtime smoke remains manual and non-CI-gated.
• Real OpenAI usage requires user-installed OpenAI SDK and OPENAI_API_KEY.
• vision://... refs are not adapter-executable yet.

Publishing status

• GitHub pre-release only.
• Package version: 0.0.2a0.

v0.0.1-alpha (MVP RC)

Highlights

• Playwright explicit-selector quickstart path for deterministic first-run smoke usage.
• Playwright natural-language act, verify, and extract usage paths for MVP workflows.
• Mobile channel routing supports act, verify, and extract via Appium adapter wiring.
• Appium quickstart is provided as a real-infrastructure template.
• Deterministic benchmark baselines:
  • Static validation: 12/12
  • Execute validation: 12/12

Known limitations

• Appium quickstart requires real mobile infrastructure:
  • running Appium server
  • running emulator/device
  • installed target app
  • local capability alignment
• Playwright quickstart is deterministic local smoke using page.set_content(...), not full real-app coverage.
• Tier 3 AI/LLM/vision/OCR behavior remains optional and depends on explicit configuration.
• PyPI/TestPyPI publishing is deferred for this MVP RC.
• Packaging currently builds successfully, but setuptools may emit non-blocking license metadata deprecation warnings.

Validation baseline

• python scripts/validate_package.py
• python scripts/validate_package.py --strict
• python -m build
• python scripts/run_benchmarks.py
• python scripts/run_benchmarks.py --execute
• pytest --collect-only -q

Status

Pre-release. Not recommended as a stable production dependency yet.