Hand-offs without GitHub: bot kit zip, bot zip import, local refresh
- python -m collection export-bot-kit: policy.json, solver.js, graph.js and a README with checksums, zipped for the bot - the bot import takes a folder or a zip (files may sit under any top folder) - python -m collection refresh: the daily workflow, run locally - trajectories files are named by date and raw checksum, so two imports on one day never overwrite each other Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UCzTd31GphpMXDhQqWrCMQ
This commit is contained in:
co-authored by
Claude Opus 5.5
parent
19209a2ad6
commit
5c49591dae
@@ -73,6 +73,12 @@ This writes `data/bot/scenes.json` and `data/bot/trajectories-<date>.jsonl`: sce
|
||||
bot's attempts to `runs_clean.csv` (source `bot`). The importer refuses to write anything that looks like prose.
|
||||
See `docs/BOT_DATA_PROTOCOL.md` for what the bot should log.
|
||||
|
||||
## Running everything locally
|
||||
|
||||
`python -m collection refresh` does what the daily GitHub workflow does: checks every source, rebuilds the runs,
|
||||
the graph, the route model and the statistics. Add `--bot <folder or zip>` to import bot runs, and
|
||||
`python -m collection export-bot-kit` zips the files the bot uses (see `docs/BOT_INTEGRATION.md`).
|
||||
|
||||
## Updating with new form data by hand
|
||||
|
||||
Download the responses spreadsheet (Google Sheets → File → Download → Microsoft Excel), then:
|
||||
|
||||
Reference in New Issue
Block a user