AgentsView

About

AgentsView brings your AI coding sessions into one searchable archive. A background server watches the session directories your agents already write and imports supported formats (over 60 agent formats are listed) into a local SQLite archive, so you can browse recorded conversations, compare activity and costs, and reuse lessons from past work. The archive stays on your machine unless you turn on a feature that shares it.

The Activity dashboard shows when agents are actually working, including peak concurrency, active versus idle time, and agent-minutes across parallel sessions, filterable by project, agent, and machine. Token and cost reports read from the pre-indexed archive, with cache-aware accounting and pricing based on LiteLLM and OpenRouter rates. Search covers full-text matching and opt-in semantic search, and tool usage and quality signals link back to the transcript for review. Recall (experimental) extracts decisions, warnings, and project facts from transcripts, with entries linked to supporting messages.

The same archive is available to agents through a CLI, a REST API, and an MCP server, and the app ships as a single Go binary with desktop, web, and CLI interfaces. SSH sync can fetch sessions from other Linux and macOS hosts, and the local SQLite archive can be pushed to PostgreSQL or ClickHouse for team use.