Gigai Chronicle

Your AI development history, inside your own repo: the prompts, the AI's responses, the tools it ran and the code it produced — recorded as plain text, linked to git and replayable. git blame tells you who wrote the line; chronicle why tells you what was asked. Local-first, zero network, MIT.

Git answers what changed. It has never answered what was asked. As AI writes more of your code, that second history — the prompts, the attempts, the reasoning — is scattered across tool-specific logs or thrown away. Chronicle keeps it, inside your own repo, travelling by git clone like everything else.

The CLI answers `chronicle why <file>` with the prompts behind a file's changes, replays whole sessions, versions the prompts that worked, and restores code to the moment before any prompt — always safety-checkpointed. `chronicle knowledge` surfaces the decisions and TODOs buried in those sessions, and `chronicle context` assembles a paste-ready brief so the next AI session starts with what was already asked and decided — both rule-based and model-free. `chronicle doctor` is the trust anchor: log integrity, a secret audit, and proof that nothing leaves your machine.

The VS Code extension (also for Cursor and Windsurf) adds sidebar history, a git-style graph of prompt versions, one-click restore and 'Why is this file like this?' in the editor. Claude Code is captured live and Codex history can be imported; Gemini and Cursor are on the roadmap. MIT licensed, published on four platforms: npm, VS Code Marketplace, Open VSX and GitHub.