# Scripting

CORE Script Execution is always available. Its optional Analysis workspace can be shown or hidden from **Modules** without disabling the backend.

Use the one Script Library to create or import a GNU Octave `.m` or Python `.py` script. Every script has a language, stable ID, version, revision, SHA-256 hash, description, required public APIs and optional typed parameters. The user library stores reusable scripts across Sessions. Assign a user script to the current Session to create a physical Session-owned snapshot. The stable reference is `user:<id>` or `session:<id>`; a filesystem path is not the Session script identity.

Run a registered script directly from the Script Library, Analysis Script Workspace or Session Script Explorer. The execution uses an exact source snapshot, validates typed parameters and writes its report, log and provenance under `analysis/scripts/` in the Session. Session-owned scripts and their executions appear in the offline HTML export; unrelated user-library scripts do not.

For the normal **Octave Script** or **Python Script** Analysis module, choose `registered_script` and select a matching-language `session:<script_id>` entry. The selector lists current Session scripts and shows language, stable ID, name, version, revision, SHA-256 and required APIs. Enable **Run / recalculate now** and run through the normal Run page. GX-TXT verifies the Session-owned source hash, executes an immutable snapshot, and records the same identity in provenance. A later user-library revision never replaces the Session snapshot. Editing the Session script keeps its ID while increasing its revision and changing its SHA-256; earlier runs remain intact. Save and reopen retain the Analysis reference.

Octave retains an `external_file` compatibility mode for standalone `.m` paths. Registered Session scripts are the normal reproducible workflow for both languages. Workspace Interactive Documents can be exported and imported into Script Library; registered scripts need no export to run in Analysis.

Python uses the configured `runtime.python`, including whatever scientific libraries the researcher installed there. Import the public `gxtxt` client for API discovery, required-API checks, context and calls. NumPy arrays are available when NumPy is installed; large numeric arrays reuse the Bridge binary transport. A missing import or Python exception produces its traceback. Read the public Parallel Execution resource policy before creating worker pools. `gxtxt_context()["output_dir"]` is a framework-owned area for scientific output files; publish completed files through `DATASET_FILE_PUBLICATION_API_v1`. See the [Python client reference](../api/PYTHON_SCRIPT_CLIENT_v1.md) for examples and transport details.

See [Script Execution API](../api/SCRIPT_EXECUTION_API_v1.md), [Sessions](sessions.md) and the [Configuration API](../api/CONFIGURATION_API_v1.md).

To package an exact `.m` or `.py` research script as an ordinary installable Analysis module, use the [Script Module Builder](script_module_builder.md) from the Modules page.