convez.ai
Agente
drimchansky

agents-kit

drimchansky/agents-kit

Ver no GitHub
InicialPluginScore · de 0 a 100

A personal kit for working with Claude Code, Codex, and other coding agents

Instalação

$npx convez-ai drimchansky/agents-kit

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
1
Forks
0
Acompanhando
0
Saúde
28%
Último commit
há 1 dia
Versão
—
Issues abertas
2
Contribuidores
1

README

A tradução para o português deste README ainda está na fila.

Agents Kit

A personal kit for working with Claude Code, Codex, and other coding agents.

Getting started

Install the kit into Claude Code and Codex with the setup script:

git clone git@github.com:drimchansky/agents-kit.git ~/agents-kit
~/agents-kit/setup.sh

Skills

Skills are organized into two groups: an engineering workflow (11 skills) that shapes the development loop, and a set of utilities (5 skills) for ad-hoc tasks.

Engineering workflow

The workflow runs roughly in order, but you don't need every step — pick whichever fits the task.

  • explore — Understand existing code or context before changing it. Example: /explore how does the retry queue work?
  • refine-idea — Sharpen a rough idea before planning — assumptions, MVP scope, "Not Doing" list. Example: /refine-idea add a draft mode to the editor
  • resume-task — Brief on, resume, or hand off a task — read the resolved task directory and report in chat. Example: /resume-task auth-jwt-migration
  • plan-task — A change is non-trivial and needs a contract. Writes paired <task-slug>.spec.md (acceptance criteria) and <task-slug>.plan.md (steps) in the resolved task directory. Example: /plan-task migrate auth to JWT
  • review-task — Confirm the implementation direction is right and still in sync with CONTEXT.md, the spec, and the current codebase; surface any drift between plan assumptions and code reality. Example: /review-task auth-jwt-migration
  • implement-plan — A validated plan is ready to ship. Marks steps [x], writes a *.result.md, and runs an acceptance gate against the spec before flipping the plan to done. Example: /implement-plan auth-jwt-migration
  • review-commit — Staged changes need a sanity check before commit — correctness, completeness, accidental inclusions, pattern fit; also drafts the commit message. Example: /review-commit
  • review-pr — A branch needs review against its base — bugs, blast radius, pattern fit, with PR context pulled from GitHub when available. Example: /review-pr
  • audit — Audit an existing module or whole project for structure, patterns, and health — no diff, no recent-change focus. Example: /audit src/auth or /audit (whole project)
  • verify-issue — A reported bug needs to be confirmed and root-caused before a fix. Example: /verify-issue users see 500 on signup
  • review-docs — Audit a doc against the codebase and surface stale references, gaps, and drift. Applies fixes only when you ask after seeing the review. Example: /review-docs README

Task directories

refine-idea, plan-task, review-task, and implement-plan share a directory-based contract that lets them hand off cleanly; resume-task reads from the same directory but never mutates it. A standalone effort lives in .agents/tasks/<slug>/ with a shared CONTEXT.md, paired *.spec.md + *.plan.md files, and append-only *.result.md records. The spec carries the acceptance criteria; implement-plan runs an acceptance gate against it before flipping the plan to done. Related efforts can be grouped under a project directory with a shared PROJECT.md (linked from each task's CONTEXT.md via a **Project:** header), and completed or skipped tasks can be moved into an archive/ subdirectory; see references/workflow/task-layout.md for the layout and discovery rules.

Reference checklists

The kit ships domain checklists under references/, partitioned by subdirectory: references/engineering/ holds technical/code-domain checklists (typescript, react, css, html, forms, interactions, tanstack-query, security, privacy, performance, accessibility, testing, code-style) preloaded by implement-plan (the skill that writes production code) and consulted ad-hoc by other engineering skills as needed. references/workflow/ holds methodology checklists (acceptance-criteria, task-lifecycle, task-layout, review) consulted by orchestration skills (plan-task, review-task, implement-plan, review-commit, review-pr, audit, resume-task, refine-idea).

Utilities

Standalone skills that aren't tied to the engineering loop:

  • proofread — Polishing a message, email, or piece of writing for grammar, clarity, and factual accuracy. Example: /proofread (paste text)
  • translate — Moving content between languages while preserving tone and context. Example: /translate to Spanish
  • fact-check — Verifying factual claims against trustworthy live sources on the internet — not against pretraining. Example: /fact-check (paste claim)
  • refine-idea-chat — Chat-only sibling of refine-idea: sharpen a rough idea through diverge/converge/sharpen phases without writing anything to disk. Example: /refine-idea-chat add a draft mode to the editor
  • review-note — Reviewing a personal knowledge base note on any subject — surfaces inaccuracies, gaps, and clarifications that deepen understanding. Proposes fixes; doesn't edit. Example: /review-note notes/stoicism.md
Ver o original no GitHub

Bom para

Relacionados