In September 2026 a hands-on report by Leiphone poured cold water on the plugin ecosystem: the ten most-downloaded DSH plugins do not boot in a single configuration — the #2, #3 and #6 plugins fight over the same UI slot, and the runtime has no arbiter to tell the user what happened.
The demand
- Leiphone’s field report on DSH plugin shortcomings (2026-09): three tiers of file permissions do nothing against plugins, top plugins conflict over slots with no arbiter, and installing a plugin silently invalidates the KV-cache prefix. Not user grumbling — measured reporting.
- A concrete case on the dshfind forum: dsh-agent-teams breaks on desktop 0.1.2-rc.1 because a removed API (registerContinuableSetup) was still in use.
- In the comments of official Discussion #5120, plugin authors share the same pain: “typecheck is green, then it explodes at runtime.”
The gap
No dedicated compatibility tool exists. LX2000WASD/dsh-web-plugin-manager (68★) bundles conflict analysis as a side feature of a do-everything manager. The runtime’s design philosophy is composition-over-arbitration, so “health check before install” will likely stay a community niche for a long time.
Proposed shape
A host service / diagnostic tool: read the active cordis composition (every installed plugin’s registration declarations) and statically detect three conflict classes — slot contention (multiple plugins claiming one UI position), service-name clashes, and version/dshTarget matrix mismatches. Output: “will this configuration boot, where will it break, and what to change.”
Implementation notes
- Composition semantics: the official capability-seams.md and slots.md.
- Read-only. Never auto-edit the user’s configuration — trust is the entire product.
- Every warning carries its evidence chain: which declaration in which plugin causes which conflict. A warning without evidence is noise.
The MVP
CLI form: point it at a profile, print the slot/service conflict list. UI in v2.
Risks
The hard part is not detection but keeping up: the harness changes weekly, and composition semantics changes force rule updates. This is a “run alongside the ecosystem long-term” project — suited to an author who wants to go deep on the plugin ecosystem itself.
FAQ
Why doesn’t the official runtime arbitrate?
Judging by the architecture docs, composition correctness is left to declaration and convention; an official arbiter is unlikely soon — which is exactly the community’s space.
Is this the same as a security scanner?
No. A scanner asks “what can this plugin reach”; a compatibility checker asks “will these plugins fight each other.” Both are needed, and the latter currently has zero supply.