{
  "markdown": "# Skills\n\nA collection of agent skills.\n\nMost large skills use a thin `SKILL.md` entrypoint plus topic packs under `references/`. Load the entry file first, then only the relevant reference files for the current task.\n\n## Install\n\n### Global (all projects)\n\n```bash\n# Install all skills globally\nnpx skills add zzci/skills --global\n\n# Install a specific skill globally\nnpx skills add zzci/skills --skill pma --global\nnpx skills add zzci/skills --skill pma-cr --global\nnpx skills add zzci/skills --skill pma-web --global\nnpx skills add zzci/skills --skill pma-rust --global\nnpx skills add zzci/skills --skill pma-go --global\nnpx skills add zzci/skills --skill pma-bun --global\nnpx skills add zzci/skills --skill react-spectrum-s2 --global\nnpx skills add zzci/skills --skill spectrum-audit --global\nnpx skills add zzci/skills --skill bkd --global\nnpx skills add zzci/skills --skill gitea --global\n```\n\n### Project (current project only)\n\n```bash\n# Install all skills to current project\nnpx skills add zzci/skills\n\n# Install a specific skill to current project\nnpx skills add zzci/skills --skill pma\nnpx skills add zzci/skills --skill pma-cr\nnpx skills add zzci/skills --skill pma-web\nnpx skills add zzci/skills --skill pma-rust\nnpx skills add zzci/skills --skill pma-go\nnpx skills add zzci/skills --skill pma-bun\nnpx skills add zzci/skills --skill react-spectrum-s2\nnpx skills add zzci/skills --skill spectrum-audit\nnpx skills add zzci/skills --skill bkd\nnpx skills add zzci/skills --skill gitea\n\n# List available skills\nnpx skills add zzci/skills --list\n```\n\n## Add PMA\n\nFirst-time PMA setup in a project:\n\n1. Ensure `AGENTS.md` has a `## Project Development` section that references `/pma` and project-specific docs/paths. Use a Chinese heading only if the user explicitly requests it.\n2. Create `CLAUDE.md` as a symlink to `AGENTS.md` so project instructions do not drift.\n3. Ensure `docs/task/index.md` exists.\n4. Ensure `docs/plan/index.md` exists.\n5. Ensure core docs exist: `docs/architecture.md` and `docs/changelog.md`.\n\n## Add PMA CR\n\nFor stack-aware review after implementation and before merge:\n\n1. Use `/pma-cr` for local diff review.\n2. Use `/pma-cr <PR-number-or-URL>` for pull request review.\n3. Use `/pma-cr audit`, `/pma-cr repo`, or `/pma-cr --repo` for repository-wide audit.\n4. The skill applies shared review policy plus only the relevant stack packs:\n   - TypeScript frontend\n   - TypeScript backend / Bun\n   - Go\n   - Rust\n   - Python\n5. Repository audit reports findings by `P0` to `P3`, plus coverage gaps and recommended next actions.\n6. Repository audit explicitly checks isolated dead code such as orphan modules, dead handlers, stale feature-flag paths, and unreachable subsystems.\n7. Dead-code output is split into `Dead Code Findings`, `Dead Code Removal Candidates`, and `Needs Runtime Verification`.\n8. Keep findings focused on correctness, security, regressions, and operational risk.\n9. Prefer repository-specific guidance from `CLAUDE.md` and `AGENTS.md` when present.\n\n## Add PMA-Web\n\nFor frontend projects using the PMA-Web stack:\n\n1. Use `/pma` for investigate -> proposal -> implement workflow.\n2. Default to a single React 19 + TypeScript + Vite 8 SPA; promote to a Bun monorepo only when multiple apps or shared packages exist.\n3. Standardize app structure around `src/app`, `src/features`, and `src/shared`.\n4. Use TanStack Router, TanStack Query, Zustand, Tailwind CSS v4, and shadcn/ui `base-nova` with `@base-ui/react`.\n5. Configure required quality gates: lint, typecheck, build, and test.\n6. Do not introduce Radix UI or other UI component ecosystems.\n\n## Add PMA-Rust\n\nFor Rust multi-crate workspace projects:\n\n1. Use `/pma` for investigate -> proposal -> implement workflow.\n2. Set up Cargo workspace defaults in `[workspace.dependencies]`.\n3. Pin a stable toolchain via `rust-toolchain.toml` and reproducible rustflags.\n4. Configure strict quality gates: fmt, cranky, deny, test, and release build.\n5. Choose data access strategy: SQLx (Default), SeaORM (Alternative), or diesel-async when compile-time schema typing is paramount.\n6. Standardize on figment + clap, Tokio, Axum, and rustls-based networking.\n7. Set up OpenTelemetry observability when deploying production services.\n\n## Add PMA-Go\n\nFor Go service and CLI projects:\n\n1. Use `/pma` for investigate -> proposal -> implement workflow.\n2. Use `cmd/` + `internal/` project layout; avoid `/pkg`.\n3. Configure strict quality gates: goimports, golangci-lint v2, go vet, test, build, mod tidy.\n4. Choose data access strategy: sqlc + pgx (Default) or GORM (Alternative).\n5. Set up layered config with koanf: defaults -> file -> env -> flags.\n6. Standardize on stdlib net/http + Chi (Default) or Gin (Alternative).\n7. Set up OpenTelemetry observability when deploying production services.\n\n## Add PMA-Bun\n\nFor Bun backend and full-stack projects:\n\n1. Use `/pma` for investigate -> proposal -> implement workflow.\n2. Default to a single API project; promote to a Bun monorepo only when multiple deployable apps or shared packages exist.\n3. Use backend modules under `src/modules` and shared code under `src/shared`.\n4. Configure strict quality gates: lint, typecheck, build, test, coverage, and security review.\n5. Choose data access strategy: Drizzle + SQLite-first storage (Default) or PostgreSQL/libSQL for multi-instance requirements.\n6. Validate environment configuration with Zod at startup and centralize runtime path resolution.\n7. Standardize on OpenAPIHono on top of `Bun.serve()`, with `app.ts` / `index.ts` / `dev.ts` split by runtime role.\n8. Use a dedicated compile pipeline when shipping standalone binaries with embedded assets or migrations.\n\n## Add BKD\n\nFor operating BKD kanban boards via REST API:\n\n1. Use `/bkd` for single issue CRUD, cron jobs, and basic operations.\n2. Use short activation phrases such as `use bkd to start coordination` or `start BKD L1` to start the three-tier L1/L2/L3 coordination pattern.\n3. For long-running autonomous coordination, L1 has no cron and wakes only on user messages or L2 follow-ups. Every campaign is split across multiple bounded L2 coordinators; each L2 owns its workstream DAG and self-cron, while L3 issues execute short-lived subtasks.\n4. For multi-subtask orchestration, the skill supports two modes:\n   - **Worktree mode**: subtasks work in isolated branches (`bkd/{issueId}`), suitable for multi-file changes or overlapping subtasks.\n   - **Simple mode**: subtasks work directly on the main branch, suitable for small independent changes.\n5. Pipeline-style quality assessment: each subtask is evaluated immediately on completion via logs filter, then self-review and coordinator assessment.\n6. Worktree mode includes branch merge strategies with post-merge build/test verification.\n7. Reference packs are loaded on demand: `rest-api.md`, `orchestration.md`, `quality-review.md`, `merge-strategy.md`, `three-tier-coordination.md`.\n\n## Available Skills\n\nUpstream provenance and refresh instructions for the Adobe skills are documented\nin [React Spectrum Skills Source](docs/sources/react-spectrum.md).\n\n| Skill | Description |\n|-------|-------------|\n| [pma](skills/pma/) | Project development lifecycle — complexity-tiered three-phase workflow + task tracking + docs + multi-agent |\n| [pma-cr](skills/pma-cr/) | Stack-aware code review — shared policy plus TypeScript frontend, TypeScript backend/Bun, Go, Rust, and Python review packs for local diffs, PRs, and repository audits |\n| [pma-web](skills/pma-web/) | Frontend implementation guide — PMA-managed React 19 + TypeScript + Vite 8 SPA conventions, TanStack Router/Query, Tailwind CSS v4, shadcn/ui base-nova, and quality gates |\n| [pma-rust](skills/pma-rust/) | Rust implementation guide — PMA-managed workspace conventions, strict quality gates, SQLx/SeaORM/diesel-async data access, Axum/Tokio patterns, OpenTelemetry, and rustls-only security defaults |\n| [pma-go](skills/pma-go/) | Go implementation guide — PMA-managed service/CLI conventions, strict quality gates, sqlc + pgx/GORM, Chi/Gin HTTP patterns, koanf config, OpenTelemetry, and slog logging |\n| [pma-bun](skills/pma-bun/) | Bun implementation guide — PMA-managed backend service conventions, optional monorepo promotion, `src/modules` API layout, SQLite-first Drizzle patterns, OpenAPIHono/Bun.serve runtime split, compile-time embedded assets, and validated env config |\n| [react-spectrum-s2](skills/react-spectrum-s2/) | React Spectrum S2 implementation guide — component selection, usage, styling, testing, and bundled React Aria references |\n| [spectrum-audit](skills/spectrum-audit/) | Spectrum adherence audit — deterministic checks, scoring, and report generation for React Spectrum S2 projects |\n| [bkd](skills/bkd/) | BKD kanban board operator — REST API workflows for projects, issues, cron jobs, event-driven L1 plus multiple cron-driven L2 workstreams and L3 execution, multi-subtask orchestration with worktree/simple modes, logs filter quality assessment, and branch merge strategies |\n| [gitea](skills/gitea/) | Gitea REST API operator — curl-based `/api/v1` reference for repos, files, issues, PRs, releases, labels, milestones, actions, packages, wiki, search, server setup, and forced non-GitHub to Gitea routing |\n",
  "bytes": 9226,
  "sha": "5d76c7525034b0f873f44b6527c12cdad732676aaff967f8e49e79101b3dc14b",
  "repo_slug": "zzci/skills",
  "fonte": "repo",
  "truncated": false,
  "api": "https://api.agentalog.com/api/listings/skl_zzci_skills_pma_rust_02031727/readme"
}