Release Mode Skills — Suite de Skills para DevRel
LeeFlannery/release-mode-skills
Skills prontas para automação completa do pipeline DevRel
Ver no GitHubConjunto portável de skills em formato SKILL.md para automação de trabalho em Developer Relations, abrangendo pesquisa de clientes, auditoria de documentação, criação de conteúdo e tracking de oportunidades. Funciona nativamente em Claude Code, Codex CLI e OpenClaw, oferecendo automação completa do pipeline de DevRel — desde prospecção até geração de métricas e conteúdo multiplataforma.
Instalação
npx convez-ai LeeFlannery/release-mode-skills1 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
Casos de uso
- Pesquisa estruturada de empresas-alvo para prospecção de clientes
- Auditoria de experiência de desenvolvedor (DX audit) em APIs e SDKs
- Geração automatizada de propostas, tutoriais e release notes
- Planejamento de apps de referência e scripts de demo
- Repurposição de conteúdo para múltiplas plataformas (YouTube, LinkedIn, X, TikTok)
- Design de frameworks de métricas DevRel com regras anti-vaidade
Veredito

Suite madura e bem-estruturada de skills para DevRel, com foco em automação prática de tarefas repetitivas (pesquisa, auditoria, escrita, tracking). Cobertura ampla do pipeline, guardrails de segurança inclusos e portabilidade garantida — ideal para equipes que gerem DevRel em escala.
README
A tradução para o português deste README ainda está na fila.
release-mode-skills
Agent skills following the open SKILL.md spec, portable across Claude Code, Codex CLI, and OpenClaw. Authored by Release Mode LLC.
Skills
A full DevRel-agency skill suite organized by pipeline stage, plus tracking, workflow, and reference skills.
Win clients
| Skill | Description |
|---|---|
| company-research | Deep research brief: org structure, where DevRel reports, product feedback loop signals |
| jd-analyzer | Score a job description against a configurable rubric |
| proposal-writer | Turn discovery-call notes into a consulting proposal or SOW |
Do the work
| Skill | Description |
|---|---|
| dx-audit | Walk an API/SDK onboarding as a new developer; friction report with time-to-first-value |
| docs-sync | Audit and update project docs to match the actual source code |
| sample-app-planner | Spec a weekend-sized reference implementation against a target API |
| tutorial-writer | Turn working code into a step-by-step tutorial with verified commands |
| demo-script | Two-column demo video script: talk track + screen actions, timestamped |
| release-notes | Turn a git log or changelog into developer-facing release notes |
Prove it
| Skill | Description |
|---|---|
| devrel-metrics | Design a metrics framework: activation, time-to-first-value, anti-vanity rules |
| case-study-writer | Turn engagement notes into a client case study — real metrics only |
Market yourself
| Skill | Description |
|---|---|
| content-repurpose | One source piece → platform-native cuts (YouTube, LinkedIn, X, TikTok, blog) |
| cfp-writer | Conference talk proposals: titles, abstract, outline, bio |
Track & organize
| Skill | Description |
|---|---|
| opportunity-tracker | Track job and contract opportunities through your pipeline |
| vault-add | Dispatcher: route an "add this" request to the right vault sub-skill |
| vault-add-saas | Add a paid SaaS / freemium / paid tool to the tools-services vault |
| vault-add-stack | Add a package / library / CLI / framework / runtime to the stack vault |
| vault-add-job | Add a job / role / application to the job-search vault |
| vault-add-client | Add a freelance / consulting lead to the potential-clients vault |
Workflow & safety
| Skill | Description |
|---|---|
| guardrails-enforcer | The Four Laws of Agent Safety — read before edit, stay in scope, verify, halt when uncertain |
| env-separator | Enforce strict test/production environment separation |
| commit-validator | Validate commits: single focus, no secrets, clean messages |
| update-todo-after-commit | Keep TODO.md synchronized with repository work after commits |
Reference
| Skill | Description |
|---|---|
| openclaw-reference | Reference for the OpenClaw agentic runtime |
| ghostty-terminfo | Fix "unknown terminal" errors when SSHing from Ghostty |
Install
The fastest way is npx skills, which works across 40+ agents (Claude Code, Codex CLI, Cursor, Gemini CLI, ...):
# pick skills interactively
npx skills add LeeFlannery/release-mode-skills
# or install specific skills, or everything
npx skills add LeeFlannery/release-mode-skills --skill dx-audit -g -y
npx skills add LeeFlannery/release-mode-skills --all
Or clone and copy whichever skills you want:
git clone https://github.com/LeeFlannery/release-mode-skills.git
cp -r release-mode-skills/skills/dx-audit ~/.claude/skills/
Restart your agent after installing so it rescans skill metadata.
Install paths by agent
| Agent | User-level | Repo-level |
|---|---|---|
| Claude Code | ~/.claude/skills/ | .claude/skills/ |
| Codex CLI | ~/.codex/skills/ | .codex/skills/ or .agents/skills/ |
| OpenClaw | ~/.openclaw/skills/ | n/a |
Quality standards
Every skill in this repo meets a bar most skill collections don't state, let alone enforce:
- Trigger-rich descriptions. Every
description:lists the concrete phrases that should activate the skill, so agents trigger it reliably instead of guessing from a one-liner. - Scope fences. Every skill says what it does not do, so it doesn't bleed into neighboring tasks or take actions you didn't ask for.
- Prompt-injection guards. Skills that consume pasted content (emails, job descriptions, changelogs, discovery notes) explicitly treat that input as data, never as instructions.
- Example outputs. Deliverable-producing skills ship a
references/example-output.mdshowing a real, complete sample of what they produce — instructions you can verify, not just trust. - CI-validated. Every push checks that frontmatter is well-formed, skill names match their folders, and README links resolve.
SKILL.md format
Every skill is a folder containing a SKILL.md file:
---
name: my-skill
description: When to trigger this skill and what it does.
---
# My Skill
Instructions in natural language.
Skills can include optional subdirectories:
references/— supporting docs loaded on demandscripts/— executable helpers for deterministic tasks
License
MIT