agentviz
Session replay visualizer for AI agent workflows (Claude Code, VS Code, Copilot CLI)
Description
Session replay visualizer for AI agent workflows (Claude Code, VS Code, Copilot CLI)
Release History
| Version | Changes | Urgency | Date |
|---|---|---|---|
| 0.7.0 | Imported from npm (0.7.0) | Low | 4/21/2026 |
| v0.7.0 | ## What's Changed - **Multi-model display**: Replay and Stats panels now show all models when a session uses multiple (e.g. \claude-opus-4.6 +2 more\) - **Duration formatting**: Fixed \ ormatDurationLong\ to handle hours (was showing \10904m\ instead of \181h 44m\) - **Stats token card**: Input/output tokens stacked on separate lines for readability - **Stats ALL MODELS section**: Replaced comma-separated text with structured rows - **Replay Session Info**: Added missing Duration row **Full Ch | High | 4/20/2026 |
| v0.6.0 | ## What's New in v0.6.0 ### Features - **Static manifest mode** -- deploy AGENTVIZ as a static site with a JSON manifest of sessions (#39) - **Copilot CLI 1.0.17+ support** -- parse JSONL sessions from the latest Copilot CLI format (#77) ### Performance - Pre-indexed search for O(1) toLowerCase on large sessions (#74) - Wrap waterfall sub-components with React.memo (#72) - Cache replay layout height estimates (#75) - Split PlaybackContext into three focused contexts (#70) ### Fixes - Multi-st | Medium | 4/12/2026 |
| v0.5.0 | ## Highlights - Dashboard landing mode with a persistent List/Dashboard toggle, shared landing controls, stronger title consistency, and low-signal discovered session demotion. - Immediate dark/light/system theme switching, including the Stats view card-surface fix. - Richer built-in demo session plus refreshed screenshots across the README and docs. - GraphView now loads ELK in a web worker, shrinking the main graph chunk dramatically, and npm v10 bin path handling is normalized. ## Validatio | Medium | 4/5/2026 |
| v0.4.0 | ## What's new ### Features - **VS Code Copilot Chat support** — parse and visualize VS Code chat session files alongside Claude Code and Copilot CLI sessions - **Dark/light/system theme switching** — full theme system with user preference persistence - **Accurate per-model cost estimation** — StatsView now shows API-reported cost alongside the model-based estimate - **Q&A drawer redesign** — Quick Insights section, persistent multi-turn conversation history, stop button, rich markdown rendering | Medium | 4/5/2026 |
| v0.3.0 | ## What's New - **Multi-agent detection** across Replay, Tracks, Waterfall, and Stats views (agent badges, colors, subagent grid) - **Fork/join DAG** in Graph view -- parallel subagents auto-fork into side-by-side branches with diamond fork/join nodes - **Temporal overlap detection** for concurrent tool calls - **Multi-agent demo** built in (`?demo=multiagent`) ## Quality - Triple code review (Claude + GPT-5.4 x2), all findings fixed - 354 tests passing - Style guide compliance across all cha | Medium | 3/31/2026 |
| v0.2.0 | **Full Changelog**: https://github.com/jayparikh/agentviz/compare/v0.1.1...v0.2.0 | Medium | 3/30/2026 |
Dependencies & License Audit
Loading dependencies...
Similar Packages
agentdashpulseAI Agent Management Dashboard — monitor and manage Claude Code, GitHub Copilot, and other AI agent sessions1.0.4
@neyugn/agent-kitsUniversal AI Agent Toolkit - Skills, Agents, and Workflows for any AI coding assistant0.5.8
@votruongdanh/ai-agent-skillsCross-IDE AI agent skills with native installers for Kiro, Cursor, Antigravity, and VS Code/Copilot3.2.1