Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
27 changes: 27 additions & 0 deletions CLAUDE.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,27 @@
# CLAUDE.md — interscript-py

The official Python runtime: deterministic transliteration over the
ISC map corpus, parsing the map DSL directly. Project purpose and
direction: `interscript-models` README.

## Facts that bite

- The engine parses `.imp`/`.isc` directly — no precompiled map
modules, no monorepo bootstrap in the runtime path.
- Release = push `v*` tag → PyPI trusted publishing. The publisher is
currently registered against the archived `secryst/secryst-py` and
must be re-registered by the owner to this repo before any release
can exchange credentials (the 0.2.1 blocker).
- Maps that need vocalized input dispatch to a crystal through the
optional phonological layer (`secryst` stdlib function); an engine
without it cannot execute maps that declare a vocalization step.
- Measured coverage is stated against maps' own embedded tests — keep
that table honest; document gaps in the suite, not in prose.

## Laws

- PRs only, never commit to main. Versions are owner decisions.
- No unconditional sibling-repo paths: dependencies resolve from the
registry; local paths are conditional fallbacks only.
- Package immutability at runtime: no writes into the installed
package.
17 changes: 17 additions & 0 deletions README.adoc
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,13 @@ This version (0.2.0) parses the `.imp` map DSL *directly* — no
pre-compiled Python map modules or monorepo bootstrap required. Point
it at the map corpus and go.

Interscript turns transliteration standards from paper tables into
executable, verifiable software: authority schemes encoded once in the
ISC language, every runtime producing provably identical output, and
neural models (diacritization, G2P) published under the same proof
regime — pinned artifacts, sha256 verification, frozen conformance
vectors.

Maps that need vocalized input (undiacritized Arabic, nikud-less Hebrew,
unsegmented Thai) dispatch to a
https://www.secryst.org[secryst crystal] through the optional phonological
Expand Down Expand Up @@ -80,6 +87,16 @@ remaining gaps are documented in the test suite:
- npm: `npm install interscript` (https://www.npmjs.com/package/interscript)
- This: `pip install interscript`

== Direction

Per-language SOTA rows land through the config-driven training spine in
https://github.com/interscript/interscript-models[interscript-models]
(a language is two YAML files; gates are machine-checkable). The
implementable contracts are the ISC map language and the IMF v1 model
format with frozen conformance vectors. See the
https://github.com/interscript/interscript-models#direction-audit-2026-10[project
direction].

== License

BSD-2-Clause. See link:LICENSE.adoc[LICENSE.adoc].
Loading