Journalism Agent
carlosceja27/journalism-agent
Co-piloto jornalístico com verificação de fontes e workflows éticos
Ver no GitHubSkill abrangente e ético para conduzir investigações jornalísticas rigorosas e bem fundamentadas em fontes. Oferece workflows estruturados de pesquisa, verificação de fontes, tratamento de alegações sensíveis e geração de relatórios com transparência editorial e conformidade ética.
Instalação
npx convez-ai carlosceja27/journalism-agent1 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
- Investigação jornalística com rastreamento de fontes
- Estruturação de reportagens baseadas em evidências verificáveis
- Pesquisa de registros públicos e documentação de denúncias
Veredito

Skill sólido e bem documentado para jornalismo investigativo com forte ênfase em ética editorial e rastreabilidade de fontes. Recomendado para redações que precisam automatizar workflows de verificação mantendo rigor jornalístico.
README
A tradução para o português deste README ainda está na fila.
Journalism Skill (V2)
A comprehensive, ethical workflow skill designed to assist journalists, researchers, and writers in conducting thorough, source-grounded investigations and reporting.
Overview
This skill provides a structured approach to journalistic research, emphasizing source transparency, strict verification, and ethical boundaries. It is designed to act as a powerful co-pilot, helping you gather facts, analyze sources, and structure reports while avoiding common pitfalls like unverified claims or placeholder citations.
Core Capabilities
- Proportional Response Depth: Delivers concise answers for simple explanatory questions, while providing full research packages only when warranted by the assignment.
- Source Transparency & Verification: Ensures that all important claims are backed by retrievable source details. Strict rules prohibit labeling any claim as "Verified" without specific, cited evidence.
- Cautious Causation: Emphasizes exact language (e.g., "amid," "following," or "in the context of") rather than inferring causation without explicit sourcing.
- Sensitive Allegations Handling: Provides robust guidelines for exact attribution, documenting denials or right-of-reply status, and handling reputation-damaging claims with extreme care.
- Robust Workflows & Checklists: Includes extensive resources, templates, and checklists for public records, interview planning, and data review.
Main Files
SKILL.md— The core skill instructions optimized for agentic environments.WEB_CHAT_PROMPT.md— A copy/paste prompt optimized for web chats (Claude Web, ChatGPT, Gemini, etc.).resources/— Deeper checklists and doctrine files.templates/— Reusable output templates for consistent reporting.examples/— Sample outputs demonstrating the V2 pattern.
How to Use
This skill is designed to be highly versatile and can be executed within both agentic environments and web chat interfaces.
Agentic Settings (CLI & Workspaces)
This skill functions best in environments where the AI agent has direct tool access (e.g., file system reading/writing, web search, URL fetching). Supported agents include Claude Code, Antigravity, Codex, Hermes, OpenClaw, and similar tools.
- Clone the repository: Clone this directory into your local workspace.
- Invoke the agent: Point your agent to the
SKILL.mdfile.- Example prompt: "Read the
SKILL.mdfile in this directory and help me structure an investigation into public records."
- Example prompt: "Read the
- Agent Workflow: The agent will autonomously read the core instructions, templates, and resource files to guide you through the reporting process. For specialized work, you can direct the agent to read specific files in the
resources/folder.
Web Chat Environments
If you are using a standard web interface without direct file system access (e.g., Claude Web, ChatGPT, Gemini), the best approach is hybrid but self-contained:
- Use the Web Prompt: Open the
WEB_CHAT_PROMPT.mdfile located in this repository. - Copy and Paste: Copy the contents of that file and paste it into your web chat interface, followed by your assignment or source materials.
- Keep it Focused: The core prompt is compact enough to guide the AI without crowding out your reporting notes.
- Optional Add-ons: Keep detailed resources from the
resources/directory as optional add-ons. Paste them into the chat only when a specific workflow matters (e.g., interview planning or anonymous sourcing).