Label compatibility evidence
Source declarations, format inference, and test results stay separate. Possible support is never presented as verified.
Platform pages go beyond filtering: they explain install paths, compatibility evidence, permission boundaries, and portability limits across clients.
Platform pages
The same SKILL.md is not automatically compatible with every agent client. Explicit declarations, static inference, and real tests are shown separately.
Agent Skills for Codex workflows. Compare installation paths, permission actions, and source evidence before adding one to your stack.
ExploreFind source-traceable Skills for Claude Code, with declared, inferred, and unverified compatibility clearly separated.
ExploreExplore Skills for Cursor project workflows, separating native support from portable workflows.
ExploreA pilot compatibility view for OpenClaw that only promotes Skills with explicit or reproducible evidence.
ExplorePage admission
Every aggregation page needs a distinct decision job, enough qualified entities, and useful content beyond a repeated card list.
Source declarations, format inference, and test results stay separate. Possible support is never presented as verified.
Show install locations, commands, triggers, credentials, and uninstall or rollback paths.
A platform needs at least 12 candidates, four qualified details, and a distinct compatibility matrix.
Continuous learning
GSC queries, internal search, saves, install copies, source clicks, and maintenance cost determine whether a page is enriched, merged, paused, or scaled.