Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

ORBIT Platform

v1.0.0

Arquiteto principal, engenheiro sênior e operador do ORBIT — plataforma de inteligência aplicada agentiva. Use quando: (1) construindo ou expandindo o aplica...

0· 118·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for drivercagropecuaria-cyber/orbit-platform.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "ORBIT Platform" (drivercagropecuaria-cyber/orbit-platform) from ClawHub.
Skill page: https://clawhub.ai/drivercagropecuaria-cyber/orbit-platform
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install orbit-platform

ClawHub CLI

Package manager switcher

npx clawhub@latest install orbit-platform
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
high confidence
!
Purpose & Capability
The skill purports to be an operational guide for the ORBIT platform and includes code snippets and explicit references to a Supabase project and the Supabase service_role key. However, the registry metadata declares no required environment variables or credentials. Requesting a service_role key and pointing to specific workspace credential files is disproportionate to an instruction-only skill unless the author explicitly requests and documents those secrets.
!
Instruction Scope
SKILL.md explicitly tells the operator/agent to read /workspace/projeto/backend/.env and /workspace/projeto/decisoes/2026-03-24-credenciais-finais.md and to 'use always the service_role key'. That directs the agent to access local secret-bearing files and to use a high-privilege DB key — behavior beyond a passive documentation skill and not declared in the skill manifest.
Install Mechanism
No install spec or code files that would be downloaded/executed are present; the skill is instruction-only, which reduces install-time risk. There is no third-party download or package installation described.
!
Credentials
Although the metadata lists no required env vars, the code examples and rules explicitly rely on SUPABASE_URL and SUPABASE_SERVICE_KEY (service_role). The skill also implicitly requires OpenAI credentials and Telegram webhook configuration. Asking for use of a service_role key (broad database privileges) without declaring it is disproportionate and risky.
Persistence & Privilege
The skill does not request permanent 'always' inclusion or modify other skills. However, because the instructions encourage using a high-privilege service_role key and reading local credential files, allowing autonomous invocation could increase blast radius. This combination amplifies risk even though 'always' is false.
What to consider before installing
This skill contains detailed operational instructions that require sensitive credentials (Supabase service_role key) and tells the agent to read local credential files, but the skill metadata doesn't declare these requirements. Before installing or using it: (1) ask the author/source to explain/declare exactly which environment variables and files are needed and why; (2) never supply a Supabase service_role key to a third-party skill — prefer a least-privileged service account or restricted RLS key; (3) avoid giving the agent access to /workspace paths that contain .env or credentials unless you trust the skill source and have audited the content; (4) run any code in a safe, isolated environment and verify the Supabase project id belongs to you or your organization; (5) if you must use the skill, rotate credentials afterward and monitor DB activity for unexpected operations. If the author can't justify the privileged access or update the manifest to declare required env vars, treat the skill as unsafe to grant secrets to.

Like a lobster shell, security has layers — review code before you run it.

agentesvk973srt4qyp8qwwevqpexgnxwx83gsstdossievk973srt4qyp8qwwevqpexgnxwx83gsstinteligencia-aplicadavk973srt4qyp8qwwevqpexgnxwx83gsstlatestvk973srt4qyp8qwwevqpexgnxwx83gsstopenaivk973srt4qyp8qwwevqpexgnxwx83gsstorbitvk973srt4qyp8qwwevqpexgnxwx83gsstpesquisavk973srt4qyp8qwwevqpexgnxwx83gsstpt-BRvk973srt4qyp8qwwevqpexgnxwx83gsstsupabasevk973srt4qyp8qwwevqpexgnxwx83gssttelegramvk973srt4qyp8qwwevqpexgnxwx83gsst
118downloads
0stars
1versions
Updated 1mo ago
v1.0.0
MIT-0

ORBIT Platform — Skill Operacional

Contexto do Projeto

ORBIT é uma plataforma agentiva de inteligência aplicada. Recebe comandos via Telegram ou Dashboard Web, aciona agentes especializados e entrega dossiês analíticos + apresentações HTML premium.

Credenciais e stack: Ver /workspace/projeto/backend/.env e /workspace/projeto/decisoes/2026-03-24-credenciais-finais.md


Stack Técnica

CamadaTecnologia
BackendNode.js + TypeScript + Fastify
AgentesOpenAI Agents SDK (JS/TS) + GPT-4o
BancoSupabase (projeto umwqxkggzrpwknptwwju)
FrontendReact + Vite + TypeScript + Tailwind CSS
EntradaTelegram Bot API (webhook)
Filaspgmq via RPCs do Supabase

Arquitetura em Camadas

Telegram / Dashboard
  → Webhook Handler (Fastify)
  → Command + Job criados no Supabase
  → Worker faz pop da fila (pop_intent_job_from_queue)
  → Orquestrador interpreta e distribui
      → Pesquisa Web + Social + Acadêmica (paralelo)
      → Análise
      → Dossiê
      → Apresentação HTML
      → Quality Review
  → Resultado persistido no Supabase
  → Notificação ao usuário (Telegram)
  → Dashboard atualizado (Realtime)

Agentes — Referência Rápida

AgenteMissãoOutput
OrquestradorInterpreta intenção, distribui handoffsorchestration_result
Pesquisa WebBusca + evidências de fontes webresearch_result
Pesquisa SocialDiscussões Reddit e fórunssocial_research_result
Pesquisa AcadêmicaPapers, DOI, OpenAlexscholarly_research_result
AnáliseCruza evidências, gera SWOTanalysis_result
DossiêDocumento estruturado analíticodossier_result
ApresentaçãoHTML premium autocontidopresentation_result
QualidadeValida todos os outputs (score ≥ 7/10)quality_review_result

Para contratos de dados completos: ver references/contratos-dados.md Para perfis detalhados dos agentes: ver references/perfis-agentes.md


Schema Supabase — Tabelas Principais

O banco (umwqxkggzrpwknptwwju) já tem 26 tabelas criadas e limpas. Não criar novas — usar as existentes.

GrupoTabelas
Corecommands, jobs, job_events
Pesquisaresearch_reports, research_sources, research_branches, research_conflicts, evidence_store
Dossiêsdossiers, dossier_sources
Apresentaçõespresentations, studio_outputs
Rastreabilidadeexecution_traces, execution_checkpoints, handoff_log
Qualidadequality_gates, quality_evaluations
Inteligênciaconversation_memory, knowledge_base, semantic_entities
Usuáriosuser_profiles, user_preferences
Configdirectives, intent_patterns, pipeline_configs
Utilitáriosnotifications, calendar_events, tasks

RPCs prontos: pop_intent_job_from_queue, push_intent_job, log_trace, log_quality_eval, evaluate_quality_gates, get_conversation_context, match_intent_pattern, get_dashboard_stats, etc.

Para schema completo com campos: ver references/schema-supabase.md


Plano de Implementação

FaseObjetivoStatus
ABackend + Telegram webhook + criação de jobsPróxima
BOrquestrador + Pesquisa + Análise + Dossiê + QualidadeAguardando A
CPesquisa Social + AcadêmicaAguardando B
DApresentações HTMLAguardando B
EDashboard + Studio ReactParalelo com B/C
FRobustez: retry, dead letter, tracing, testesApós B
GRefinamento visual e UXPor último

Padrões de Código

Supabase Client (backend)

import { createClient } from '@supabase/supabase-js'
const supabase = createClient(
  process.env.SUPABASE_URL!,
  process.env.SUPABASE_SERVICE_KEY!
)

Criar job no Supabase

const { data: job } = await supabase
  .from('jobs')
  .insert({ command_id: cmdId, chat_id: chatId, status: 'pending' })
  .select().single()

Pop da fila

const { data } = await supabase.rpc('pop_intent_job_from_queue')

Log de trace

await supabase.rpc('log_trace', {
  p_job_id: jobId, p_agent: 'orchestrator', p_version: '1.0.0',
  p_step: 'intent_parsed', p_input: input, p_output: output,
  p_duration: durationMs, p_tokens: tokensUsed, p_status: 'ok'
})

Agente OpenAI SDK

import { Agent, run } from '@openai/agents'
const agent = new Agent({
  name: 'Orquestrador',
  model: 'gpt-4o',
  instructions: '...',
  tools: [...],
  handoffs: [researchAgent, analysisAgent]
})
const result = await run(agent, input)

Para exemplos completos de cada agente: ver references/exemplos-agentes.md


Regras de Execução

  1. Nunca codifique sem entender o fluxo — verifique os contratos e o schema
  2. Use sempre a service_role key do projeto umwqxkggzrpwknptwwju
  3. Respeite as foreign keys ao deletar dados (filhos antes dos pais)
  4. Quality gate obrigatório antes de qualquer entrega — score mínimo 7/10
  5. Logs estruturados via log_trace RPC — nunca console.log em produção
  6. HTML das apresentações deve ser autocontido — sem CDN externo
  7. Idioma padrão: Português do Brasil em toda comunicação com o usuário
  8. Nunca expor credenciais no código — sempre via variáveis de ambiente

Quality Gates

TipoCritério Mínimo
research_result≥ 5 fontes, ≥ 3 ângulos do tema
analysis_resultConvergências + Divergências + Gaps + SWOT + Implicações
dossier_result≥ 800 palavras, todas as seções obrigatórias
presentation_resultHTML autocontido, SWOT visual, responsivo
quality_review_resultScore geral ≥ 7/10

Referências Adicionais

  • Contratos de dados JSON: references/contratos-dados.md
  • Perfis completos dos agentes: references/perfis-agentes.md
  • Schema Supabase completo: references/schema-supabase.md
  • Exemplos de código por agente: references/exemplos-agentes.md
  • Metodologia de trabalho: /workspace/projeto/metodologia/METODOLOGIA.md
  • Workflows reutilizáveis: /workspace/projeto/workflows/WORKFLOWS.md
  • Plano mestre: /workspace/projeto/PLANO-MESTRE.md

Comments

Loading comments...