erikas-skills

Read-only cross-artifact consistency check across spec.md, plan.md, and tasks.md — coverage gaps, orphan tasks, unsafe deferrals, incomplete story checkpoints, and unverified safeguards. Produces checklists/consistency-report.md and a READY_FOR_EXECUTION verdict; never modifies spec.md, plan.md, tasks.md, or state.json. When a spec-analyzer agent is requested, generates compatible definitions for both Kiro CLI and Claude Code.

Create or update .speckit/memory/constitution.md — a project's non-negotiable architecture, security, and quality principles — from explicit documentation and/or by mining consistent patterns across existing code. Every mined principle carries evidence, coverage, and confidence, and none becomes binding without explicit user confirmation. Optional, and re-invocable to check whether current code has drifted from an existing constitution. When a pattern-miner agent is requested, generates compatible definitions for both Kiro CLI and Claude Code.

Execute tasks.md in dependency order, producing real code — delegating implementation discipline to verify-before-implement, commit messages to git-commits, and a final safety-net review to pre-pr-review, instead of reimplementing any of their rules. Runs stories sequentially by default, or in parallel across independent stories/tasks via task-executor subagents when the story boundaries genuinely allow it. Never invents scope beyond tasks.md/plan.md/spec.md, never silently re-touches a task already marked done, and never skips a story's closure checkpoint. When a task-executor agent is requested, generates compatible definitions for both Kiro CLI and Claude Code.

Generate or update a technical implementation plan (plan.md) from an approved REASONS specification, enforcing verified project context, two-gate Constitution Check, convention enforcement, and the Safe Deferral rule for phased work; when a plan-reviewer agent is requested, generate compatible definitions for both Kiro CLI and Claude Code.

Generate or update a governed, implementation-ready specification from a PRD, document, file reference, or free-form text using SPDD and the REASONS Canvas; when support agents are requested, generate compatible definitions for both Kiro CLI and Claude Code.

Decompose an approved technical plan (plan.md) into an executable, user-story-organized task list (tasks.md) — Setup, Foundational, one phase per user story in priority order with a mandatory closure checkpoint, then Polish. Enforces the Safe Deferral rule from spec-kit-generate-plan and traceability to spec.md/plan.md; when a task-decomposer agent is requested, generates compatible definitions for both Kiro CLI and Claude Code.

Detect drift between spec.md/plan.md/tasks.md and their recorded state.json fingerprints — mechanical hash comparison plus best-effort structural localization by stable ID — and recommend which upstream spec-kit skill to re-invoke, one hop at a time. Never edits spec.md, plan.md, or tasks.md itself; that stays each generating skill's own "Updating an Existing X" procedure. Invocable standalone after editing a source document, or as the Fase 0 staleness check inside spec-kit-generate-plan, spec-kit-generate-tasks, and spec-kit-execute-tasks. When a drift-watcher agent is requested, generates compatible definitions for both Kiro CLI and Claude Code.

Document code clearly, concisely, and professionally. Use when the user asks to document functions, classes, structs, or code files; add comments to existing code; improve existing documentation; or mentions "document", "comments", "docstrings", "JSDoc", "godoc", or "code documentation". Generates documentation that other engineers can quickly understand without being verbose, with strict prose line-count caps and no comment-width limit, while allowing one line per structured parameter, return, or error entry and forbidding per-property documentation except for extreme type ambiguity.

Write clear, semantic git commit messages and PR descriptions following Conventional Commits. Use when the user asks to write commit messages, create PR descriptions, format git history, improve commit quality, or mentions "commit message", "conventional commits", "PR description", "merge message", or "git history". Generates professional, consistent commit messages that make project history easy to understand.

Analyze and implement code changes with strict verification protocol. Use when implementing features, fixing bugs, or modifying existing code. Enforces reality-checking against actual codebase, contracts, and specs before making changes. Prevents assumption-based development and invented parameters/entities.

Standard for initializing new projects by asking which architecture skills to apply and providing setup commands with latest library versions. Never generates the project directly; always provides commands for the user to execute. Ensures libraries are always at their latest versions without hardcoding version numbers.

Defines mandatory standards for REST API contracts and data format, including unified response structure, OpenAPI documentation, ULID identifiers, JWT auth headers, mandatory entity fields, and HTTP status codes. Framework-agnostic — describes what the API must return, not how to build it.

Defines a Clean Architecture and reusable component structures for frontend. Especially useful when the user wants to organize pages, tables, lists, item cards, subpages, and modal dialogs in a predictable and maintainable way. Applies to React, Vue, Svelte, or similar frameworks.

Generate academic reports, papers, and technical documentation using LaTeX. Use when the user asks to create research papers, lab reports, theses, technical documentation, IEEE-style papers, or any academic document requiring professional typesetting. Produces publication-quality PDFs with proper formatting, citations, figures, tables, and mathematical notation. Ideal for Computer Science, Engineering, and scientific documents.

Analyze and explain code from multiple levels - from implementation details to software architecture. Use when understanding new codebases, explaining system design, documenting architecture, learning patterns, or analyzing technical decisions. Provides technical deep-dives, architectural overviews, flow diagrams, and concrete examples. Works at any abstraction level from "what does this function do" to "how does this system scale".

Analyze code changes before creating a PR with production-impact focus. Use when reviewing your own changes, preparing a PR, or conducting self-review before requesting team review. Identifies breaking changes, integration bugs, missing validations, and regressions by verifying against actual codebase, contracts, and production flows. Generates detailed, actionable feedback organized by severity.