load_skill
- Python
9
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill bdambrosio/cognitive_workbench --skill load- Skill.md2.8 KB
Overview
This skill loads a persisted Note or Collection by ID or name and returns its content (optionally sliced). It supports Python-style slice strings to control how much of the Note text or how many items from a Collection are returned. The skill validates slices and returns structured, prefixed content rather than the raw resource object.
How this skill works
You call load with a target (Note or Collection ID/name), an out binding name, and an optional slice string like "0:1000" or ":". For Notes the skill returns a prefixed text string containing the requested character range and metadata; for Collections it creates a new Collection containing the requested items and returns a planner-visible preview listing each item ID and its first 200 characters. Validation rejects only clearly invalid slices (both start and stop non-negative and stop < start) and missing or non-existent targets cause failure.
When to use it
- Retrieve part or all of a persisted Note by ID or name for processing or display
- Extract a subset of items from a persisted Collection into a new Collection
- Chunk large Notes into multiple loads for incremental processing
- Obtain a preview of Collection contents for planning or UI display
- When you need a content string with count metadata rather than the resource object
Best practices
- Specify slice explicitly (use ":" for full content) when you need the entire Note or Collection beyond default ceilings
- Use chunked loads (incremental slices) to process very large Notes without hitting resource limits
- Load by stable ID when available to avoid ambiguous name resolution
- Do not attempt to load resources that are not persisted or to load from inside another Collection
- Treat empty content as valid output, not an error
Example use cases
- Load the first 4,096 characters of a long Note for summarization: load(target=Note_123, slice="0:4096", out=$chunk)
- Iteratively process a large document by loading 500-char windows: load(... slice="0:500"), load(... slice="500:1000"), etc.
- Create a smaller Collection preview by loading the first 10 items from a large Collection: load(target=Collection_45, slice="0:10", out=$mini)
- Fetch an entire Note explicitly with slice=":" when you need full content without default ceilings
- Produce a planner-visible preview of items (ID + first 200 chars) when inspecting a Collection
FAQ
The skill returns failed for missing or invalid resource IDs/names.
How do I get the full content without the default ceiling?
Provide an explicit slice of ":" or a range that covers the full content; explicit slices bypass the default ceiling.
Can I load from a Collection (i.e., pass a Collection as the source of Notes)?
No. load accepts individual Note or Collection targets only; it does not accept an enclosing Collection as a source for loading its members.