Command reference
All commands live in the repository root and read .env.
Capture
| Command | Does |
|---|---|
./sp-start [--mode portal|live] [--fg] | Start the recorder with the capture profile for the mode. Normally the app runs this |
./sp-stop | Stop the recorder and the auto-switcher |
./sp-mode [live|portal|auto] [--json] | Show, pin or unpin the mode |
./sp-domains list|add <host[:subdomains]>|remove <host> | Manage the portal allowlist |
./diag-drm [seconds] | Phased test of which capture path blanks protected video |
Lectures
| Command | Does |
|---|---|
./lecture list [--since 7d] [--json] | Sessions found in the captures |
./lecture build [<id>|last|all] [--since 7d] | Build one or more sessions |
./lecture build --from HH:MM [--to HH:MM] [--title T] | Build one session for a time window |
… --no-vision | Skip slide descriptions |
… --cards N | Upper bound on cards (default 20) |
./lecture watch [--idle 5] | Build each session a few minutes after it ends |
./lecture doctor [--quick] [--json] | Health checks |
./transcript_ref.py <folder> <reference.txt> | Merge a reference transcript and measure word error rate |
Study
| Command | Does |
|---|---|
./study ask "question" [--since 7d] [--app NAME] [--json] | Answer from your captures, with sources |
./study summarize [--since 3h | --date YYYY-MM-DD] | Notes for a period |
./study flashcards [--since 3h] [-n 20] | Cards for a period |
./study check | Recorder and model server health |
Evaluation
| Command | Does |
|---|---|
tests/evals/run.sh [--smoke|--judge-only|--regen-goldens] | Judged evaluation |
tests/evals/run_judge_bench.sh | Judge benchmark |
eval/run_eval.py --configs think_off | Slide-reading evaluation |
Maintenance
| Command | Does |
|---|---|
./check.sh | Unit tests, contract checks, self-tests |
app/build_app.sh [--test] [--open] | Build, sign and install the app |
tools/setup_hooks.sh | Turn on the content guard hooks |
tools/content_guard.py --all | Scan the tracked tree |
tools/publish.sh [--dry-run] | Guarded push |