spec

Generate a spec file for a new feature or bugfix

Break down a validated specification into actionable implementation tasks

Implement a validated specification with review-fix loops

End-to-end spec implementation with review loops - validates, decomposes, and executes sequentially

Analyzes a specification document to determine if it has enough detail for autonomous implementation

Fixes a specific code review issue. Prioritizes by severity: CRITICAL issues first (security, crashes, data loss), then IMPORTANT (performance, error handling). Called by task-executor after review identifies issues. <example> Context: Code review found a security vulnerability prompt: "Fix CRITICAL issue: SQL injection in user query at src/db/users.ts:42. Current code uses string concatenation for query building." assistant: "I'll analyze the vulnerability, implement parameterized queries, and verify the fix doesn't break existing tests." <commentary> The issue-fixer focuses on a single issue, applies the fix, and verifies it works. </commentary> </example> <example> Context: Code review found missing error handling prompt: "Fix IMPORTANT issue: Unhandled promise rejection in api/fetch.ts:28. The async call has no try-catch." assistant: "I'll add proper error handling with try-catch, ensure errors are logged, and add appropriate user-facing error messages." <commentary> Agent handles the specific issue without over-engineering. </commentary> </example>

Executes a single task from a decomposed spec with the full implementation cycle: implement, test, review, and fix. Automatically updates STM status upon completion. <example> Context: Orchestrator needs to implement a specific task from a spec prompt: "Execute task [P1.3] from STM: Implement common hook utilities" assistant: "I'll load the task details from STM, implement the code, write tests, run code review, fix any critical issues, then mark the task complete." <commentary> The task-executor handles the complete lifecycle of a single task autonomously. </commentary> </example> <example> Context: A task needs implementation with testing prompt: "Execute task [P2.1]: Create user authentication module" assistant: "Loading task details... implementing auth module... writing tests... running review... all checks passed, marking task done." <commentary> Agent follows the full cycle and auto-updates STM status. </commentary> </example>

This skill should be used when the user asks about "spec methodology", "specification workflow", "spec-driven development", "how to write specs", "spec best practices", "specification templates", or needs guidance on creating, validating, decomposing, or executing specifications for software development.

1 hook across 1 event