What problem does it solve?
Developers adding new cross-cutting behaviors to VS Code's agent-host chat lifecycle often place them directly in AgentSideEffects or AgentService, which bloats these orchestration classes and breaks established architectural boundaries. This Skill enforces the contribution model so new behaviors live in self-contained, dependency-injected units with explicit ordering and clear hook semantics.
Core Features & Use Cases
- Decision Framework: Provides a yes/no test to determine whether a new behavior belongs as a contribution versus routing fabric or provider-shaped code.
- Hook Reference: Documents the seven
IAgentHostChatContribution hooks, their firing moments, ordering constraints, and failure-isolation rules.
- Anatomy & Registration: Specifies the required shape of a contribution (unique id, explicit order, context-first constructor) and the single registration site in
builtInChatContributions.ts.
- Review Rule: Flags code added to
AgentSideEffects or AgentService that lacks a routing, correctness, provider, or dispatcher justification.
- Use Case: When reviewing a pull request that adds a new lifecycle side effect, prompt injection, restored-history transformation, or action observer, apply this Skill to verify the contribution model is followed and ordering tests are updated.
Quick Start
Review the proposed change against the agent-host chat contributions model and confirm the new behavior is registered as a contribution with explicit ordering.