Catalog vs structural inventory: how do you treat same-name skills across repos? #1063
Replies: 2 comments
|
That boundary is correct by design. The AAS Core tools |
|
Thanks — that clarifies the boundary cleanly. Takeaway on my side: AAS Core stays on the verified catalog + I won’t treat that inventory job as something AAS should absorb. If I write up a short “pre-selection inventory” note (what inputs, what collisions it surfaces, what it explicitly does not claim), I’ll keep it complementary to AAS rather than overlapping Core tools. Appreciate the correction. |
Uh oh!
There was an error while loading. Please reload this page.
Topic area
Repo feedback
What would you like to discuss?
I read the AAS Core framing around complete local catalog search, agent-owned selection,
compose_stack, and keepingaas-stack.jsonplus evidence reviewable — without ranking or certifying semantic completeness.One adjacent boundary I’d like corrected if wrong: AAS helps an agent choose and validate exact catalog IDs, while builders may still need a repo-local pass that asks “what Skill/MCP-like units are already present in these agent repos, where are their source files, and which short names collide across repos?”
Do you see cross-repo same-name discovery as part of AAS Core /
inspect_stack, or intentionally as a separate structural-inventory concern?A one-line correction would help. Thanks for making the supported vs experimental boundaries unusually explicit.
Relevant links or examples
No response
Before posting
All reactions