-
Notifications
You must be signed in to change notification settings - Fork 2
Commands
You mostly run one thing — tron start. TRON then runs autonomously; you're not in a REPL.
Wake TRON. A short bootup asks, in order: AIDE's own model (blank keeps the default), the
scope (all, a phase, or a range of blocks), the worker count, whether to pause for your OK
before each merge to trunk (ask-before-merging), and the model for each role (architect,
reviewer, worker — each with a recommended default you confirm or override). AIDE (a real LLM, counsel
only, fail-open) also recommends a next block along the way. A non-interactive boot (e.g. the SIM
harness) asks nothing and takes defaults. Then TRON dispatches the fleet and drives the pipeline to done,
printing milestone notes as blocks land. Pass a project path to skip the picker.
TRON pages you (terminal OPERATOR> prompt, or Telegram if configured) only when it needs a
decision. You reach TRON by dropping a file in the project root:
| Drop a file | Does |
|---|---|
parley.md |
a question or instruction — the architect answers, recorded under parley/
|
report-request.md |
the architect writes a status report, recorded under reports/
|
STOP |
under --watch, ends the run |
| Command | What it does |
|---|---|
tron --watch |
long-running: idles, wakes on register work (WAKE), exits on a STOP file |
tron --selftest |
the engine selftests — no agents, no tokens |
python3 engine/tron.py |
the engine directly — prompts for the project path (no demo project ships with a clone) |
(tron is the launcher on your PATH after install; from a clone it's ./tron.)
| Command | What it does |
|---|---|
python3 engine/<module>.py |
that module's selftests (gate, glossary, events, workflow, bpmn, tg) |
python3 engine/glossary.py --write |
regenerate docs/GLOSSARY.md (same for events/workflow; bpmn.py --write regenerates workflow/) |
python3 evaluation/harness.py <template> N |
run N SIMs of a template; --parallel N, --ablate ARM
|
See also: Configuration · Operations.