Hi everyone 👋 I built Claudescope because I kept losing track of what I'd done across dozens of Claude Code sessions, and the history sitting in ~/.claude is effectively write-only: impossible to read, impossible to search, and no idea what any of it cost. The part I find most useful day to day is the MCP server. Instead of only humans digging through old transcripts, the agent can query them directly — "have we hit this error before?", "what did we decide about the auth flow last time?" It cuts down a lot of re-explaining. The hardest part was that no agent documents its transcript format and they all shift between releases, so the connector layer was the only way to keep this maintainable. Contributions for new agent connectors are very welcome. I'd love feedback on what's missing for this to be something you'd open daily, and which agents to support next.

Claudescope

One local viewer for all your AI coding-agent sessions

One local viewer for all your AI coding-agent sessions