LibraryMemory Skill
Description
Checks the persistent research memory for cached digests before issuing a WebFetch.
Usage Guidance
- •When provided a URL (or a short identifier), compute its hash and look for the corresponding file in the memory directory.
- •If a digest exists, surface the summary and citations to the user and recommend skipping a fresh WebFetch unless explicit freshness is required.
- •If no digest is found, continue with normal research workflow.
Tooling Expectations
- •Use the Memory Tool (
memory.view) or client-provided helper scripts to read stored digests. - •Never overwrite user-provided knowledge-base assets; keep cached digests in the dedicated memory folder.
Example Trigger Prompt
code
Before fetching https://example.com/topic, call the library memory skill to see if we already have a digest for it.
Safety
Only read from memory; do not modify or delete entries unless explicitly instructed by the user.