ai-skills

Use when creating a new repository, initializing a project from scratch, or setting up OSS infrastructure (CI/CD, linting, releases, docs). Also use when user says "new repo", "init project", "set up repository", "repo-kickstart", or "kickstart repo".

Use when the user needs to search and download court decisions from CENDOJ (Centro de Documentación Judicial del Consejo General del Poder Judicial), the Spanish official judicial database. Also when user says "buscar sentencias", "descargar sentencia CENDOJ", "jurisprudencia CENDOJ", "ECLI", or provides a poderjudicial.es URL. STRICTLY for personal/individual use (uso particular). Bulk downloads or commercial use are explicitly forbidden by CENDOJ legal notice and require formal authorization from CGPJ.

Use when wiring a local development service (dev server, preview server, static site, any long-running HTTP upstream) to portless so it gets a stable `.localhost` URL instead of a port number. Also use when user says "portless", "replace ports with names", "stable localhost URL", ".localhost domain", "portless proxy", "PORTLESS_*" env vars, "sin puerto", or when integrating portless into Makefile targets or package.json scripts. Covers proxy setup, naming conventions, env vars, troubleshooting (including the cert-0-bytes stuck state), and points to ecosystem-specific references (Python, Node/Vite/Storybook, Go) loaded on demand.

Use when starting a new backend or fullstack project, evaluating an existing project's architecture, migrating from code-first to contract-first, introducing Clean Architecture layers, setting up a type generation pipeline, or assessing architectural maturity. Also use when user says "contract-first", "clean architecture", "schema SSOT", "type generation", "ports and adapters", "hexagonal architecture", "architecture assessment", "dependency inversion", "composition root", "architecture evaluation", "contract-first clean arch", "screaming architecture", "vertical slicing", "criteria pattern", or "bounded contexts".

Use when generating or editing draw.io diagrams with UML shapes. Provides correct shape-to-style mappings for deployment, component, sequence, state machine, package, and class diagrams. Also use when asking about "draw.io shapes", "UML diagram styles", "drawio XML generation", "diagram shape reference", or "UML notation in draw.io".

Evaluate any artifact (document, architecture, infrastructure design, domain model, code, PR, proposal) using an independent panel of LLM agent evaluators. Each evaluator runs as a separate agent with its own context — no cross-contamination between evaluators. Produces individual evaluation reports + a consolidated panel report. Use when the user says "expert panel", "panel de expertos", "evaluate with experts", "multi-agent evaluation", "jury evaluation", "independent review panel", "evaluate this report/design/architecture", or when objective, multi-perspective evaluation of any artifact is needed. Also use when the user wants to assess vendor deliverables, review proposals, or validate technical decisions with simulated expert input.

Interact with Huly project management platform via its TypeScript/WebSocket API. Create and manage issues, milestones, components, documents, and attachments programmatically. Use when syncing work between local project state and Huly tracker/wiki. Handles auth, class IDs, gotchas, and content upload (markdown, files). Covers both Cloud and self-hosted instances.

Use when creating or editing a Makefile that orchestrates long-running local services (dev servers, preview servers, build+serve workflows). Also use when user says "make a Makefile", "service orchestration", "standard targets", "make start/stop/restart", "Makefile convention", "make serve", "dev server Makefile", or when structuring a project with .env.local profiles and start/stop lifecycle. Defines canonical target names, namespacing with `/`, help-as-default-goal with ANSI colors, integration with .env.local, and PROFILE/APP variable patterns.

Use when adding background service management to a project that runs dev servers, preview servers, or any long-running processes. Also use when user says "service manager", "background server", "start/stop services", "PID management", "log rotation", or needs Makefile targets for start/stop/status/tail/logs.

Use when validating, auditing, or reviewing Agent Skills for compliance with the agentskills.io specification. Also use when creating new skills to verify format requirements, when asked to "validate skill", "check skill", "audit skill", "lint skill", or "review skill compliance".