Releases: Coratch/runwhy
Releases · Coratch/runwhy
v0.5.0 - React web UI migration
Changes
- Ships the React/Vite Web UI as the default shell while keeping
--ui classicas a fallback. - Adds long-session turn window APIs, full-session turn search, session search, and Markdown session export from the server.
- Normalizes Claude Code reasoning, tool calls, tool results, hooks, and attachments into separate timeline turns.
- Preserves recap and analysis workflows in the new React UI with focused inspector, search, and responsive layouts.
- Includes packaged
web/distassets and a lockfile for reproducible web builds.
Verification
- npm test: 50/50 passed
- npm run check: passed, including TypeScript and sensitive information scan
- npm run web:build: passed
- npm run pack:check: passed, 45 files in package
- Real local service verified on http://127.0.0.1:4502 without demo mode
v0.4.2 - Codex session display
Changes
- Refined Codex session display around user, assistant, tool calls, tool results, and compacted context records.
- Hid noisy Codex event messages from the session timeline while preserving task completion data for recap generation.
- Added Codex duplicate-message handling and AGENTS.md context filtering so session titles and user turns stay focused on real user input.
- Simplified the session toolbar to a single turn type filter: ALL, USER, Assistant, TOOL_CALL, TOOL_RESULT, COMPACTED.
- Added regression coverage for Codex duplicate events, compacted records, export behavior, and UI contract.
Verification
- npm test: 36/36 passed
- npm run check: passed
- git diff --check: passed
- Real local service verified on http://127.0.0.1:4500
v0.4.1 - Repository rename and README screenshot
Changes
- Renamed the GitHub repository to Coratch/runwhy.
- Updated repository homepage to https://coratch.github.io/runwhy/.
- Updated README and static site links to the new repository and portal.
- Added a fresh RunWhy Web UI screenshot to README and the static portal.
- Published runwhy@0.4.1 with updated repository and homepage metadata.
Verification
- npm test
- npm run check
- npm run pack:check
v0.4.0 - RunWhy rebrand
Highlights
- Rebranded the product from AgentLens to RunWhy.
- Published the primary npm package as runwhy@0.4.0.
- Added the new runwhy CLI while keeping agent-session-viewer as a compatibility alias.
- Repositioned the product as a local flight recorder for AI coding agents.
- Added a GitHub Pages static landing site and a static example Web page under docs/.
- Updated README, demo docs, package metadata, and frontend branding.
Install
npm install -g runwhy
runwhy --demoCompatibility
The previous npm package @coratch/agent-session-viewer remains available for existing users, but new installs should use runwhy.
Verification
- npm test: 35 passed
- npm run check: passed, including sensitive information scan
- npm run pack:check with temporary npm cache: passed for runwhy@0.4.0
- Chrome headless screenshots checked for the local app, landing page, and static example page.
v0.3.1 - Top bar layout fix
Fixes
- Fixed the AgentLens top bar being clipped after the layout refresh.
- Reset the product title styling in the app bar so the header no longer overflows the fixed top region.
- Added responsive wrapping for provider/search/refresh/days/recap controls.
- Fixed narrow-screen grid placement so the sidebar and main workspace stack instead of creating an implicit second column.
- Moved the detail-rail toggle to the lower-right on narrow screens to avoid covering top controls.
Verification
- npm test: 35 passed
- npm run check: passed, including sensitive information scan
- npm run pack:check with temporary npm cache: passed for @coratch/agent-session-viewer@0.3.1
- Chrome headless screenshots checked at 1024px, 768px, 430px, and 360px widths.
v0.3.0 - AgentLens time diagnosis
Highlights
- Renamed the product experience to AgentLens while keeping the npm package and CLI name compatible as @coratch/agent-session-viewer.
- Added single-session LLM time diagnosis through local Claude CLI print mode (claude -p), with trusted local duration metrics, delay timeline, confidence, root causes, and evidence turn jumps.
- Added persisted browser-side analysis reports, Markdown copy, and report clearing.
- Reworked the app layout: top-level source refresh/search/provider controls, session-focused left rail, independent collapsible right inspector, and session reload beside the title.
- Added Markdown session export, stricter redaction options, issue/PR recap example, and post-release metrics docs.
Verification
- npm test: 35 passed
- npm run check: passed, including sensitive information scan
- npm run pack:check: passed for @coratch/agent-session-viewer@0.3.0
Publishing note
- GitHub release is published from tag v0.3.0.
- npm publish is pending because this machine is not logged in to npm (npm whoami returns ENEEDAUTH).
v0.1.2
Local session replay for Claude Code and Codex JSONL logs.\n\nHighlights:\n- Add packaged demo mode: agent-session-viewer --demo\n- Add sanitized Claude Code and Codex JSONL fixtures under examples/fixtures/\n- Add screenshot walkthrough in docs/demo.md\n- Add README comparison against adjacent Claude Code and Codex tools\n\nInstall:\nbash\nnpm install -g @coratch/agent-session-viewer\nagent-session-viewer --demo\n