Avaliação de superfície pública — Academia Lendária
A página de marketing (hospedada em Vercel, stack Next.js) está bem cuidada: autenticação de e-mail forte (DMARC com rejeição, SPF com falha dura, DKIM Google), HSTS, arquivo security.txt, e nenhum segredo no JavaScri…
Resumo executivo
Porque agir agora
Estes achados não são teóricos: são superfícies públicas que um humano ou um agente hacker pode consumir hoje. Cada cartão tem prova, impacto, dono, fix e validação.
A página de marketing (hospedada em Vercel, stack Next.js) está bem cuidada: autenticação de e-mail forte (DMARC com rejeição, SPF com falha dura, DKIM Google), HSTS, arquivo security.txt, e nenhum segredo no JavaScript público. O risco real é o DNS da marca expondo sistemas de operação e produto — em especial Grafana com métricas abertas sem login, SSH aberto no servidor do portal e um provedor de autenticação de personificação (impersonation) anunciado publicamente.
Stack nesta corrida
| Componente | Usado | Nota |
|---|---|---|
| ParviClaw Core + AWP | SIM (lab) | pack completo no engagement |
| ParviSight browser | parcial | HTTP walk / agentic |
| PayBotFin Witness live | conforme eng | honestidade no anexo |
Achados
Cada achado: problema, onde (link), impacto, dono, PoC, fix, validar.
Grafana com métricas públicas (sem autenticação)
ALTOhttps://academialendaria.ai/ — 200, ~480 KB; version in /api/health last_reprobe=2026-07-31.
Public engineering/tooling yields free recon (versions, configs, APIs).
- Human: Public eng surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Probe APIs/health in loops; correlate versions with CVEs.
- If invasive/destructive: Source/secrets/CI risk if auth weak later. L3 only proves public exposure.
Academia Lendária — segurança / produto / infra (confirmar dono interno)
Read-only. Does not change the system.
curl -sSI 'https://academialendaria.ai/'
Remove from public DNS or require VPN/SSO; disable anonymous APIs; re-test from the public internet on academialendaria.ai.
Remove from public DNS or require VPN/SSO; disable anonymous APIs; re-test from the public internet on academialendaria.ai.
Porta SSH 22 aberta no servidor do portal
ALTOhttps://academialendaria.ai/ (host IP) + TCP/22 — OpenSSH 9.6 banner on 207.180.223.38 last_reprobe=2026-07-31.
SSH on public web IP is a permanent brute-force target.
academialendaria.ai · TCP/22
- Human: SSH on public IP is a permanent brute-force target.
- Agent (LLM+tools): Port-scan brand IPs; spray against TCP/22.
- If invasive/destructive: Shell on host. L3 only proves port open.
Academia Lendária — segurança / produto / infra (confirmar dono interno)
Read-only. Does not change the system.
nc -zv -w 3 academialendaria.ai 22
Close TCP/22 on public edge or restrict by allowlist/bastion; keys-only; fail2ban; no password auth.
Close TCP/22 on public edge or restrict by allowlist/bastion; keys-only; fail2ban; no password auth.
Provedor de personificação (impersonation) anunciado no Auth.js
ALTOhttps://academialendaria.ai/ — /api/auth/providers JSON last_reprobe=2026-07-31.
Public auth/federation edges expand phishing and IdP attack surface.
- Human: Public auth surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Keep surface in continuous recon.
- If invasive/destructive: Escalate if second bug/credential. L3 public-max is read-only.
Academia Lendária — segurança / produto / infra (confirmar dono interno)
Read-only. Does not change the system.
curl -sSI 'https://academialendaria.ai/'
Remove non-prod auth edges from public DNS; prefer auth code+PKCE; conditional access; minimize public metadata where possible.
Remove non-prod auth edges from public DNS; prefer auth code+PKCE; conditional access; minimize public metadata where possible.
Interface do Grafana no DNS da marca
ALTOhttps://academialendaria.ai/ — ** Anyone on the internet can read Prometheus-style metrics and Grafana health/version without login. last_reprobe=2026-07-31.
Public engineering/tooling yields free recon (versions, configs, APIs).
- Human: Public eng surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Probe APIs/health in loops; correlate versions with CVEs.
- If invasive/destructive: Source/secrets/CI risk if auth weak later. L3 only proves public exposure.
Academia Lendária — segurança / produto / infra (confirmar dono interno)
Read-only. Does not change the system.
curl -sSI 'https://academialendaria.ai/'
Remove from public DNS or require VPN/SSO; disable anonymous APIs; re-test from the public internet on academialendaria.ai.
Remove from public DNS or require VPN/SSO; disable anonymous APIs; re-test from the public internet on academialendaria.ai.
CSP só em modo relatório, com scripts inseguros
MÉDIOEvidence: CSP-Report-Only (or equivalent) still permissive for GTM/Facebook style stacks. Safe PoC: curl -sSI https://www.academialendaria.ai/ | grep -i content-security Fix: Move to enforcing CSP; reduce unsafe-inline/eval over time. --- last_reprobe=2026-07-31.
Weak browser security headers increase impact of XSS/clickjacking/MITM.
- Human: Public headers surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Keep surface in continuous recon.
- If invasive/destructive: Escalate if second bug/credential. L3 public-max is read-only.
Academia Lendária — segurança / produto / infra (confirmar dono)
Read-only. Does not change the system.
curl -sSI 'https://www.academialendaria.ai/' | egrep -i 'strict-transport|content-security|x-frame|x-content'
Add/tighten HSTS, CSP, XFO, nosniff, Referrer-Policy, Permissions-Policy on https://academialendaria.ai/.
Add/tighten HSTS, CSP, XFO, nosniff, Referrer-Policy, Permissions-Policy on https://academialendaria.ai/.
Cabeçalho CORS liberado para qualquer origem (*)
MÉDIOAll www responses advertise <code>Access-Control-Allow-Origin: *</code>. Lower risk on static pages, noisy and dangerous if authenticated APIs share the origin. last_reprobe=2026-07-31.
Over-permissive CORS can enable cross-origin data theft.
- Human: Public cors surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Keep surface in continuous recon.
- If invasive/destructive: Escalate if second bug/credential. L3 public-max is read-only.
Academia Lendária — segurança / produto / infra (confirmar dono)
Read-only. Does not change the system.
curl -sSI 'https://academialendaria.ai/'
Reflect only trusted Origins; avoid Access-Control-Allow-Origin: * on credentialed APIs.
Reflect only trusted Origins; avoid Access-Control-Allow-Origin: * on credentialed APIs.
Certificado TLS compartilhado com igorrover.com.br
MÉDIOEvidence: CSP-Report-Only (or equivalent) still permissive for GTM/Facebook style stacks. Safe PoC: curl -sSI https://www.academialendaria.ai/ | grep -i content-security Fix: Move to enforcing CSP; reduce unsafe-inline/eval over time. --- last_reprobe=2026-07-31.
Weak browser security headers increase impact of XSS/clickjacking/MITM.
- Human: Public headers surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Keep surface in continuous recon.
- If invasive/destructive: Escalate if second bug/credential. L3 public-max is read-only.
Academia Lendária — segurança / produto / infra (confirmar dono)
Read-only. Does not change the system.
curl -sSI 'https://www.academialendaria.ai/' | egrep -i 'strict-transport|content-security|x-frame|x-content'
Add/tighten HSTS, CSP, XFO, nosniff, Referrer-Policy, Permissions-Policy on https://academialendaria.ai/.
Add/tighten HSTS, CSP, XFO, nosniff, Referrer-Policy, Permissions-Policy on https://academialendaria.ai/.
Endpoint de saúde do portal público
MÉDIO<code>/api/health</code> returns anonymous JSON with schema status — useful for recon. last_reprobe=2026-07-31.
Public metrics leak ops data useful for targeting.
- Human: Public metrics surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Keep surface in continuous recon.
- If invasive/destructive: Escalate if second bug/credential. L3 public-max is read-only.
Academia Lendária — segurança / produto / infra (confirmar dono)
Read-only. Does not change the system.
curl -sSI 'https://academialendaria.ai/'
Require auth on metrics/Grafana/Prometheus or put behind VPN; never expose raw metrics publicly.
Require auth on metrics/Grafana/Prometheus or put behind VPN; never expose raw metrics publicly.
Hostname Prometheus público (autenticação básica)
MÉDIO<code>prometheus.academialendaria.ai</code> asks for user/password (better than Grafana) but remains brand-DNS attack surface. last_reprobe=2026-07-31.
Public engineering/tooling yields free recon (versions, configs, APIs).
- Human: Public eng surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Probe APIs/health in loops; correlate versions with CVEs.
- If invasive/destructive: Source/secrets/CI risk if auth weak later. L3 only proves public exposure.
Academia Lendária — segurança / produto / infra (confirmar dono)
Read-only. Does not change the system.
curl -sSI 'https://prometheus.academialendaria.ai'
Remove from public DNS or require VPN/SSO; disable anonymous APIs; re-test from the public internet on academialendaria.ai.
Remove from public DNS or require VPN/SSO; disable anonymous APIs; re-test from the public internet on academialendaria.ai.
Registros CAA confusos / malformados
MÉDIOMultiple certificate issuers and malformed values (e.g. duplicated “issuewild” text). last_reprobe=2026-07-31.
Public brand surface on academialendaria.ai expands recon until closed.
- Human: Public edge surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Keep surface in continuous recon.
- If invasive/destructive: Escalate if second bug/credential. L3 public-max is read-only.
Academia Lendária — segurança / produto / infra (confirmar dono)
Read-only. Does not change the system.
curl -sSI 'https://academialendaria.ai/'
Remove unnecessary public hosts on academialendaria.ai DNS or enforce strong auth/WAF; re-test from internet.
Remove unnecessary public hosts on academialendaria.ai DNS or enforce strong auth/WAF; re-test from internet.
Caminhos desconhecidos devolvem erro 502
MÉDIOWhat: Auth.js exposes credentials provider id impersonation on public API. Where: https://portal.academialendaria.ai/api/auth/providers → 200 Body includes: "impersonation":{"id":"impersonation","name":"Credentials","type":"credentials",...} Safe PoC: ``bash curl -sS https://portal.academialendaria.ai/api/auth/providers | jq . `` Expected if fixed: Provider list wi last_reprobe=2026-07-31.
Public auth/federation edges expand phishing and IdP attack surface.
- Human: Public auth surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Keep surface in continuous recon.
- If invasive/destructive: Escalate if second bug/credential. L3 public-max is read-only.
Academia Lendária — segurança / produto / infra (confirmar dono)
Read-only. Does not change the system.
curl -sSI 'https://portal.academialendaria.ai/api/auth/providers'
Remove non-prod auth edges from public DNS; prefer auth code+PKCE; conditional access; minimize public metadata where possible.
Remove non-prod auth edges from public DNS; prefer auth code+PKCE; conditional access; minimize public metadata where possible.
HSTS do domínio raiz mais fraco que o www
MÉDIOSafe PoC: ``bash curl -sS -D- -o /dev/null -H "Origin: https://evil.example" https://www.academialendaria.ai/ | grep -i access-control ` Observed access-control-allow-origin: '' Impact: Cross-origin read of responses that set ACAO ; usually lower on static marketing, higher if APIs inherit. Fix: Restrict ACAO to known origins; never *` with credentials. --- last_reprobe=2026-07-31.
Weak browser security headers increase impact of XSS/clickjacking/MITM.
- Human: Public headers surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Keep surface in continuous recon.
- If invasive/destructive: Escalate if second bug/credential. L3 public-max is read-only.
Academia Lendária — segurança / produto / infra (confirmar dono)
Read-only. Does not change the system.
curl -sSI 'https://evil.example' | egrep -i 'strict-transport|content-security|x-frame|x-content'
Add/tighten HSTS, CSP, XFO, nosniff, Referrer-Policy, Permissions-Policy on https://academialendaria.ai/.
Add/tighten HSTS, CSP, XFO, nosniff, Referrer-Policy, Permissions-Policy on https://academialendaria.ai/.
DNS da marca apontando para n8n
MÉDIO<code>n8n.academialendaria.ai</code> resolves; HTTPS timed out this run (possible firewall). Still public inventory. last_reprobe=2026-07-31.
Public engineering/tooling yields free recon (versions, configs, APIs).
- Human: Public eng surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Probe APIs/health in loops; correlate versions with CVEs.
- If invasive/destructive: Source/secrets/CI risk if auth weak later. L3 only proves public exposure.
Academia Lendária — segurança / produto / infra (confirmar dono)
Read-only. Does not change the system.
curl -sSI 'https://n8n.academialendaria.ai'
Remove from public DNS or require VPN/SSO; disable anonymous APIs; re-test from the public internet on academialendaria.ai.
Remove from public DNS or require VPN/SSO; disable anonymous APIs; re-test from the public internet on academialendaria.ai.
Supabase liberado no connect-src da CSP
MÉDIOBrowser may talk to <code>*.supabase.co</code>. Ensure only intended anon key and RLS — no service key in client. last_reprobe=2026-07-31.
Weak browser security headers increase impact of XSS/clickjacking/MITM.
- Human: Public headers surface on academialendaria.ai expands attacker recon.
- Agent (LLM+tools): Keep surface in continuous recon.
- If invasive/destructive: Escalate if second bug/credential. L3 public-max is read-only.
Academia Lendária — segurança / produto / infra (confirmar dono)
Read-only. Does not change the system.
curl -sSI 'https://academialendaria.ai/' | egrep -i 'strict-transport|content-security|x-frame|x-content'
Add/tighten HSTS, CSP, XFO, nosniff, Referrer-Policy, Permissions-Policy on https://academialendaria.ai/.
Add/tighten HSTS, CSP, XFO, nosniff, Referrer-Policy, Permissions-Policy on https://academialendaria.ai/.
Anexos de evidência (AWP + Core)
Pack real do retest L3 max 2026-07-31 (public-max, não destrutivo). Cada ficheiro tem legenda. Links abrem o artefacto.
O que é: recibo Agent Witness Protocol assinado (verificável offline).
O que prova: integridade do registo do nosso teste/engagement.
O que NÃO é: não prova sozinho o bug no vosso host (isso é o PoC de cada achado).
→ descarregar awp-receipt.json
O que é: leaves append-only (scope, probes, catálogo, quality, seal).
O que prova: timeline desta corrida de retest.
→ descarregar leaves.jsonl · leaf-chain.txt (PASS)
O que é: status/headers/samples dos probes public-max.
→ descarregar retest-raw.json
O que é: resumo multi-página (URLs, tamanhos, forms).
→ descarregar agentic-walk.json · core-export.json
→ abrir pasta evidence · engagement eng_academialendaria_l3_retest_1361 · witness_mode local_awp_dev (chaves DEV_LAB — não cerimónia de produção).
Como usar / verificar o AWP (open-source)
- Descarregue
awp-receipt.jsondo anexo (ou a pasta evidence). - Em qualquer máquina com Node 18+:
# na pasta evidence npx agent-witness-protocol verify awp-receipt.json
O que o PASS significa (honesto): assinatura, inclusão Merkle e checkpoint do recibo estão corretos — o log do teste está íntegro.
PASS NÃO significa: que o bug em produção está corrigido, nem que houve PayBotFin Witness live (este pack é local_awp_dev / DEV_LAB salvo indicação).
PayBotFin Witness: emit de rede é opcional e por engagement. Se não constar network_emit, não reivindique verify live PayBot nesta corrida.
Ordem de correção para o time
- Fechar exposições ALTO primeiro (auth / VPN / DNS / política).
- Corrigir MÉDIO (headers, e-mail, non-prod).
- Retestar cada PoC na internet pública até fechar.
- Arquivar pack AWP/evidência para auditoria e retest.