Generates TypeScript code skeletons with typed stubs and canonical TODO(implementation) placeholders.
npx skills add https://github.com/alvis/.agents --skill draft-code-alvis
Please help me install this Agent Skill. Skill: draft-code Source: https://github.com/alvis/.agents/tree/main/plugins/coding/skills/draft-code Command: npx skills add https://github.com/alvis/.agents --skill draft-code-alvis
TODO(implementation): markers so downstream completion steps know exactly what to implement. ## Core Features & Use Cases - Typed skeleton generation: Creates interfaces, type aliases, enums, and function stubs with JSDoc, marking every incomplete body with TODO(implementation): comments and an IMPLEMENTATION: sentinel throw. - Test scaffolding: Drafts pending runtime tests with describe.todo()/it.todo() and compiler-semantic type tests through the project's configured type-test mechanism. - Standards enforcement: Applies naming, file-structure, function, testing, and TypeScript standards, then verifies with tsc --noEmit, lint, and repository-native test commands. - Use Case: Given an instruction like "Create user authentication service with login, logout, and token refresh", it produces types.ts, auth.service.ts stubs, and auth.spec.ts scaffolds ready for coding:complete-code and coding:complete-test. ## Quick Start Ask the agent to draft a TypeScript skeleton for a specific feature, for example: draft a REST endpoint skeleton for product CRUD operations with typed stubs and test scaffolds.High-intent search queries and answers about installing and using this skill.
Provide a specific instruction naming the feature or module, such as a service with its methods. The skill derives interfaces, function signatures, and file structure, then writes stubs marked with TODO(implementation) comments and verifies they compile with tsc --noEmit.
Use `// TODO(implementation): <explicit behavior>` comments plus a thrown `IMPLEMENTATION:` sentinel error wherever a value is expected. Plain TODO markers are rejected because downstream completion skills only claim the canonical forms.
It drafts test scaffolds too: describe.todo()/it.todo() while no runtime entrypoint is callable, minimal failing assertions once one is, and type tests via the project's configured mechanism for compiler-observable behavior. Declaration shape alone receives no test scaffold.
Avoid it when the instruction is too vague to derive types, when you need actual business logic implemented, or when modifying existing code. Those cases belong to implementation, refactoring, or fix workflows instead.
Rejection happens when the instruction lacks enough detail to derive meaningful types, the target directory does not exist, or the skeleton would conflict with existing code structure. Verification failures from tsc or lint are fixed and re-run before reporting.
dfrysingerMap approved designs to files, interfaces, and architectural diagrams.
ResgridStructure .NET 10 xUnit and integration tests with WebApplicationFactory and Testcontainers.
QBall-IncAnalyze code for security vulnerabilities, type safety, and linting violations.
HonigbartWrite executable plans to .ai/plans/<slug>/final_plan.md.