Why STAN Over Alternatives?

STAN focuses on a single goal: making AI‑assisted development reproducible and auditable today, using your existing toolchain.

Tool / Approach	Reproducible Context	AI‑Ready Archives	Unified Diff Patching	Deterministic Outputs	Learning Curve
STAN	Yes (archive.tar + outputs)	Yes (attach in any chat)	Yes (with FEEDBACK loop)	Yes (stdout/stderr captured)	Low
“Just chat + copy/paste”	No	No	Manual/error‑prone	No	Low
git stash / branches	Partial (source only)	No	Manual	No	Medium
Nix/Guix	Strong builds	No (not chat‑oriented)	N/A	Indirect	High
IDE context managers	Varies	Varies	Varies	No	Medium

Notes:

STAN complements git, CI, and your existing scripts. It doesn’t replace them.
You control exactly what the assistant reads; archives and outputs are explicit and portable.
The FEEDBACK handshake gives you robust, self‑identifying patch correction.

Use STAN when:

You want a simple, CLI‑first workflow that works with any AI chat client.
You care about auditability and deterministic inputs, not speculative memory.
You want unified diffs and a disciplined patch loop instead of “magic changes.”