Antigravity

Antigravity keeps separate local histories for its CLI, the Antigravity 2.0 app, and the IDE. Import all three source IDs to cover every place you use it.

Source IDs

Source	Use it for
`antigravity-cli`	Antigravity CLI transcript logs, conversation databases, task files, and usage data.
`antigravity`	Antigravity 2.0 readable records, conversation databases, task files, and usage data.
`antigravity-ide`	Antigravity IDE transcript logs, conversation databases, task files, and usage data.

How the import works

Hansard reads each of the three Antigravity histories separately. The CLI source reads terminal transcript logs and task, plan, and walkthrough files. The antigravity source reads the records Antigravity 2.0 saves, and the IDE source reads editor logs and a database for each conversation. Each source uses the session IDs Antigravity recorded to rebuild messages, tool calls and results, status notices, the working directory, and any model and token data.

When a readable transcript log exists, Hansard uses it for the conversation. Without one, Hansard rebuilds the conversation from the step types it can read in the conversation database: your messages and the agent’s responses, tool calls and results, errors, notifications, and subagent launches. The database also supplies the model and token counts for each response. Step types Hansard does not recognize are counted and kept in the archive’s copy of the original files. Sessions from the CLI, the app, and the IDE always stay separate conversations.

Older conversations that Antigravity left only as conversations/*.pb files, summaries, or task files can be recovered through the running app when it still serves them. With Antigravity open, run:

hansard import antigravity --recover --since all

Hansard asks the app’s local language server for each conversation it lists and for every conversation Hansard knows about only from those older files, because the app lists a conversation only after it has loaded it. Add --session <id> to recover specific conversations. Recovery only reads, stays on your computer, keeps each response in the archive, and never stores the app’s access token. A readable transcript log or a complete conversation database still takes precedence, and later imports keep the recovered conversation.

hansard import --sources antigravity-cli,antigravity,antigravity-ide --since all
hansard index rebuild

With Automatic source selection, the watcher already includes all three sources. With a Custom list, enable them. The watcher picks them up within 30 seconds:

hansard config sources enable antigravity-cli,antigravity,antigravity-ide

Review the import in the app

When the import finishes, open Sessions and selectFilter & sort.
Set Source to Antigravity CLI, Antigravity 2.0, or Antigravity IDE and setTime to the window you imported. UseFolder to narrow the list when you use this harness in several projects.
Select a conversation to check its messages, tool calls, files, commits, model, and token usage.

To keep archiving new sessions from a local harness, openSettings → Imports. Automaticfollows every supported harness installed on this computer.Custom lets you turn individual harnesses on or off. Cloud sources turn on separately under Cloud agents, and ChatGPT, Claude.ai, Gemini, and Gemini Notebook / NotebookLM exports import withImport export…. Each also has its own import command. If a conversation you expect is missing after you reset the filters, run the import command on this page again with --explain-skips where the source supports it.

What Hansard preserves

Transcript messages, tool calls and results, status notices, task files, and exit codes.
Subagent runs, each linked to its child conversation.
System, checkpoint, history, generated-message, and error notices.
Conversation database steps, including tool activity, when no readable transcript log exists.
Recorded response models and input, output, cache-write, cache-read, and reasoning counters from the conversation databases. Reasoning is already included in output tokens.
Model-selection changes as settings history, separate from recorded response models. A selection alone does not identify which model answered.
Generation-start times when present, without treating them as completion times.
Copies of the conversation databases with their -wal and -shm companion files, plus matching annotation files when they exist.

Known limitations

Older conversations/*.pb files are kept in the archive but not decoded directly. Their conversations are readable only through hansard import antigravity --recover while the app is open, and only when the app can still return them; a conversation it cannot return is reported and keeps its existing archive.
A recovered conversation stays partial when the app returns fewer steps than it reports or includes step types Hansard does not recognize.
Database step types Hansard does not recognize are counted and kept, but do not appear as messages.
A tool call with no saved result shows no result. A nonzero exit code counts as a failure, even when Antigravity reported a general success status.
A transcript log takes precedence over database messages. A result absent from that log can remain only in the preserved database.
Transcript-log replay uses preserved logs and generated messages. Replay of database-only and language-server captures remains unavailable.
Native checks covered retained CLI history. The inspected app history was empty, and no IDE store or legacy conversation protobuf corpus was present. This does not establish history completeness on other devices or accounts.

Refresh an existing archive

Reimport earlier archives to correct token accounting, remove model attribution inferred from settings changes, and restore tool usage and generic results for database-only conversations:

hansard import --sources antigravity-cli,antigravity,antigravity-ide --since all
hansard index rebuild