Choose a DASCore workflow

Identify the installed APIs, inspect data carefully, and choose a recipe for the scientific task.

Keywords

dascore, workflow, agents, documentation

Start with the DASCore installed in the project’s Python environment. dascore doc reports its version, Python executable, package location, and documentation cache. Use that reported executable with -m dascore for subsequent commands.

Find the current API

Read installed signatures and docstrings before writing calls from memory:

python -m dascore doc Patch.select

Substitute the selected environment’s executable for python. When the operation’s name is unfamiliar, search the Markdown directory reported by doc using your editor or agent’s file-search tools. Search API names, keyword metadata, and scientific terms such as filter, bandpass, inventory, or geometry, then read the relevant sections. Point the search explicitly at that directory, since the user cache may be outside the project workspace. The local documentation guide includes ripgrep examples.

DASCore’s main objects serve different roles:

• A Patch holds an array with coordinates, dimensions, units, and metadata.
• A Spool selects, groups, and iterates over patches, including data whose arrays have not yet been read.
• An Inventory describes the observing system, its resources, geometry, configuration, and changes over time.
• An AnnotationSet describes events and regions against explicit time and distance reference systems.

Inspect before materializing

For unfamiliar files, inspect format and metadata with the installed scanning and spool APIs before reading full arrays. Check coverage, coordinate spacing, units, gaps, and overlaps. Estimate array memory from known shapes and dtypes, allowing room for intermediate arrays; if the metadata is insufficient, report that uncertainty and inspect a bounded preview.

A lazy spool can still trigger large reads when iterated or materialized. Choose explicit time and distance bounds, then check the representative patch before expanding the workflow.

Choose the procedure

python -m dascore skills
python -m dascore skill make-inventory

The skills catalog selects ordinary documentation recipes. Use make-inventory to create or repair observing-system metadata, validate it, round-trip its serialization, and test attachment to representative data.

For other scientific workflows, search the documentation and state the assumptions that affect processing: physical quantity and units, coordinate semantics, sampling, gaps, and chunk boundaries. Validate a small case before applying the workflow to an archive. Keep uncertain deployment facts and scientific interpretations explicit.