Releases: Neilblaze/portscope
Releases · Neilblaze/portscope
v1.8.3
Added
- Dependabot Automation — Configured
.github/dependabot.ymlfor weekly ecosystem package updates and automatic PR generation for both the core CLI tool and the React website. - Website CI Checks — Added a
test-websitejob to.github/workflows/pr-checks.ymlthat automatically lints and builds the React website on every pull request. - Explicit Cancellation Option — Added a
0(Exit/Cancel) option to the model selection prompt in interactive mode, allowing users to back out gracefully without selecting a model. - Live Pricing Synchronization — Enhanced the
/usagedashboard to auto-sync LLM pricing data seamlessly in the background. The CLI now performs an unawaited, asynchronous fetch to pull the latestllm-pricing.json(maintained every 12h via GitHub Actions) directly from the repository if the local user cache is older than 24 hours. This ensures cost estimates remain completely accurate for existing users without requiring a CLI update.
Changed
- Node 18 CI Deprecation — Dropped Node 18 from the GitHub Actions (
pr-checks.yml) test matrix to focus CI test cycles exclusively on active LTS (Node 20) and current (Node 22) environments. - PNPM CI Migration — Fully migrated GitHub Actions pipelines (
pr-checks.yml,coverage.yml,deploy-website.yml,release.yml) fromnpmtopnpmusing native cache configuration for significantly faster builds. - Model Discovery Strict Filtering — Overhauled the model fetching logic (
src/config/models.js) with a strictcleanModelListsanitizer. Actively filters out non-chat capabilities (e.g.,transcribe,tts,diarize,search,image, and experimentalossvariants) ensuring only relevant conversational models are displayed in the CLI. AddedmodelsUrltoschema.jsfor OpenAI and Gemini for dynamic discovery. - Advanced Pricing Resolution — Rewrote cost estimation logic in
src/ai/usage.jsto ingest the dynamically syncedsrc/data/llm-pricing.json. Replaced basic fallback mapping with an intelligentgetRates()resolver that automatically accounts for provider prefixes and model variations. - UI Aesthetics & Alignment — Upgraded the provider-flow prompts to use rich
chalk.bgGreen(✔) andchalk.bgRed(✕) block icons for action feedback. Fixed terminal spacing and alignment for API key outputs. Adjusted ASCII banner spacing. - Test Coverage & Error Styling — Expanded native Node.js test suite with robust UI formatting (
src/ui/format.js) and error sanitation (src/config/sanitize-error.js) coverage. Replaced basicsanitizeErrorglobally with styledformatChatError. Reached 296 total tests with 100% test coverage in targeted files, pushing overall line coverage past 50.90%.
Fixed
- Documentation Links — Fixed malformed paths to the
CONTRIBUTING.mdandAI_USAGE_POLICY.mdfiles acrossREADME.mdandCODE_OF_CONDUCT.md.
v1.8.2
Full Changelog: v1.8.0...v1.8.2