Narrative walkthroughs that bind APTWatcher's primitives to concrete operator stories. Each scenario is reproducible from a cold checkout with the commands and inputs it ships.
The docs/scenarios/ tree carries the reference rubric scenarios
(S01–S03). This directory (scenarios/ at the repo root) is for demo
scenarios: scripts used to walk through a capability end-to-end at a
pitch session, a status review, or an internal rehearsal. They are
narrative-first, not rubric-first.
| ID | Title | Demonstrates |
|---|---|---|
| S04 | Offline to online bundle handoff | core.bundle export + import, signed Ed25519 handoff, aptwatcher analyze --sign and aptwatcher publish against the stub adapter |
| S05 | Find evil in one shot (E01 to signed PDF bundle) | Cold-workstation one-shot triage: aptwatcher run against an E01 + memory dump on a 14.5-minute budget, ending in a signed IncidentBundle and a judge-readable report |
Each scenario file follows the same outline:
- Story — the narrative beat in one paragraph.
- Environment — offline / online hosts, transport assumptions.
- Inputs — synthetic findings + IOCs shipped with the scenario.
- Commands — exact CLI invocations, matched to the argparse
surface of
src/agent_extension/*.py. - Audit events — the names that land in the offline and online audit logs at each step.
- Success rubric — pass / partial / fail bands.
- Related — cross-links into design notes, KB procedures, and source.
S<NN>-<kebab-slug>.md. NN is zero-padded, kebab-slug is the
one-line headline. Pick a slug that will still read cleanly a year
from now (e.g. offline-to-online-handoff, not phase-3-7-demo).
../docs/scenarios/README.md— rubric scenarios S01–S03 (accuracy measurement, not demo scripts).../docs/design/— design notes each demo scenario exercises.../knowledge/procedures/— operator procedures referenced by demo scenarios.