Autonomous OODA-loop pentesting agent. Hermes brain, Kali + Playwright tools over MCP, ATT&CK as the Orient language.
Python ai-agentsautonomous-agentskali-linuxllm-agentsoffensive-securityooda-looppentestingplaywrightsecurity
docs(ctf): translate pentest report example to English
irraz8d484f368f8c
diff --git a/labs/looney-tunables-ctf/docs/pentest-report-example.html b/labs/looney-tunables-ctf/docs/pentest-report-example.htmlindex 65e69f7..0ac5d92 100644--- a/labs/looney-tunables-ctf/docs/pentest-report-example.html+++ b/labs/looney-tunables-ctf/docs/pentest-report-example.html@@ -1,5 +1,5 @@<!DOCTYPE html>-<html lang="es">+<html lang="en"><head><meta charset="UTF-8"><meta name="viewport" content="width=device-width, initial-scale=1.0">@@ -213,7 +213,7 @@white-space: pre;line-height: 1.7;}- .terminal-body .cmd { color: #e8e8e8; }+ .terminal-body .cmd { color: #e8e8e8; }.terminal-body .out-ok { color: var(--green); }.terminal-body .out-err { color: var(--red-bright); }.terminal-body .comment { color: var(--text-muted); }@@ -316,10 +316,10 @@<div class="report-meta"><div class="meta-item"><span class="meta-label">Engagement</span><span class="meta-value">eng-ubuntu-server-01-lpe</span></div><div class="meta-item"><span class="meta-label">Target</span><span class="meta-value">ubuntu-server-01</span></div>- <div class="meta-item"><span class="meta-label">Modo</span><span class="meta-value">pentest</span></div>- <div class="meta-item"><span class="meta-label">Acceso inicial</span><span class="meta-value">SSH agent_user (uid=1001)</span></div>- <div class="meta-item"><span class="meta-label">Sistema</span><span class="meta-value">Ubuntu 23.04 lunar · Docker · kernel 6.2.0-39-generic</span></div>- <div class="meta-item"><span class="meta-label">Fecha</span><span class="meta-value">2026-10-18</span></div>+ <div class="meta-item"><span class="meta-label">Mode</span><span class="meta-value">pentest</span></div>+ <div class="meta-item"><span class="meta-label">Initial access</span><span class="meta-value">SSH agent_user (uid=1001)</span></div>+ <div class="meta-item"><span class="meta-label">System</span><span class="meta-value">Ubuntu 23.04 lunar · Docker · kernel 6.2.0-39-generic</span></div>+ <div class="meta-item"><span class="meta-label">Date</span><span class="meta-value">2026-10-18</span></div></div><div class="container">@@ -338,188 +338,186 @@<span class="badge badge-high">LPE</span><span class="badge badge-high">CISA KEV</span><span class="badge badge-info">T1068</span>- <span class="badge badge-medium">Exploit público</span>+ <span class="badge badge-medium">Public exploit</span></div></div><div style="text-align:right"><div style="font-size:11px;color:var(--text-muted);margin-bottom:4px">OODA LOOP</div>- <div style="font-size:28px;color:var(--green);font-family:monospace;font-weight:bold">14 pasos</div>+ <div style="font-size:28px;color:var(--green);font-family:monospace;font-weight:bold">14 steps</div><div style="font-size:11px;color:var(--text-muted)">OBSERVE→ORIENT→DECIDE→ACT→REPORT</div></div></div>- <!-- ── 1. Resumen ejecutivo ── -->+ <!-- ── 1. Executive Summary ── --><div class="section">- <div class="section-title"><span class="section-num">01 /</span> Resumen ejecutivo</div>+ <div class="section-title"><span class="section-num">01 /</span> Executive Summary</div><div class="exec-summary">- Se identificó y demostró una vulnerabilidad de escalada de privilegios local (LPE) de severidad- <strong>alta</strong>: <strong>CVE-2023-4911</strong> ("Looney Tunables") en la glibc del sistema.- El paquete <code>libc6</code> está instalado en la versión <code>2.37-0ubuntu2</code>- (fijada con <code>apt hold</code>, estado dpkg <code>hi</code>), una revisión por detrás del- parche de seguridad oficial <code>2.37-0ubuntu2.1</code> (USN-6409-1).<br><br>- Cualquier usuario local —incluido <code>agent_user</code> sin privilegios— puede ejecutar código- como root a través de binarios SUID estándar (<code>su</code>, <code>sudo</code>,- <code>mount</code>, <code>passwd</code>…). En este engagement (<strong>modo pentest</strong>) se- demostró la explotabilidad con un crash confirmado y <strong>100% reproducible</strong> del- <code>ld.so</code> privilegiado; no se obtuvo shell root, conforme al alcance declarado.+ A <strong>high-severity</strong> local privilege escalation (LPE) vulnerability was identified+ and demonstrated: <strong>CVE-2023-4911</strong> ("Looney Tunables") in the system's glibc.+ The <code>libc6</code> package is installed at version <code>2.37-0ubuntu2</code>+ (pinned with <code>apt hold</code>, dpkg state <code>hi</code>), one revision behind the+ official security patch <code>2.37-0ubuntu2.1</code> (USN-6409-1).<br><br>+ Any local user — including the unprivileged <code>agent_user</code> — can execute code as root+ through standard SUID binaries (<code>su</code>, <code>sudo</code>, <code>mount</code>,+ <code>passwd</code>…). In this engagement (<strong>mode: pentest</strong>), exploitability was+ demonstrated with a <strong>100% reproducible</strong> confirmed crash of the privileged+ <code>ld.so</code>; no root shell was obtained, in accordance with the declared scope.</div></div>- <!-- ── 2. Hallazgo principal ── -->+ <!-- ── 2. Main Finding ── --><div class="section">- <div class="section-title"><span class="section-num">02 /</span> Hallazgo principal</div>+ <div class="section-title"><span class="section-num">02 /</span> Main Finding</div><table>- <tr><td class="td-key">CVE</td><td class="td-val"><strong>CVE-2023-4911</strong> — buffer overflow en <code>ld.so</code> (glibc) al procesar la variable de entorno <code>GLIBC_TUNABLES</code></td></tr>- <tr><td class="td-key">Técnica MITRE ATT&CK</td><td class="td-val"><code>T1068</code> — Exploitation for Privilege Escalation</td></tr>- <tr><td class="td-key">Vector</td><td class="td-val">Local: usuario sin privilegios → ejecuta binario SUID root con <code>GLIBC_TUNABLES</code> malformada → <code>ld.so</code> (corriendo como root) desborda el heap en <code>tunables_strdup()</code> durante <code>__tunables_init()</code>, antes de que el modo seguro sanee el entorno → corrupción de estructuras del loader → ejecución de código como root</td></tr>- <tr><td class="td-key">Introducida</td><td class="td-val">glibc 2.34 (abril 2021, commit <code>2ed18c5b534d9e92fc006202a5af0df6b72e7aca</code>, BZ #27471)</td></tr>- <tr><td class="td-key">Versión afectada</td><td class="td-val"><code>libc6 2.37-0ubuntu2</code> — confirmada con <code>dpkg -l</code> / <code>ldd --version</code></td></tr>- <tr><td class="td-key">Versión parcheada</td><td class="td-val"><code>libc6 2.37-0ubuntu2.1</code> — USN-6409-1 (pocket security)</td></tr>+ <tr><td class="td-key">CVE</td><td class="td-val"><strong>CVE-2023-4911</strong> — buffer overflow in <code>ld.so</code> (glibc) while processing the <code>GLIBC_TUNABLES</code> environment variable</td></tr>+ <tr><td class="td-key">MITRE ATT&CK</td><td class="td-val"><code>T1068</code> — Exploitation for Privilege Escalation</td></tr>+ <tr><td class="td-key">Attack vector</td><td class="td-val">Local: unprivileged user → runs a SUID-root binary with malformed <code>GLIBC_TUNABLES</code> → <code>ld.so</code> (running as root) overflows the heap in <code>tunables_strdup()</code> during <code>__tunables_init()</code>, before secure mode sanitises the environment → loader struct corruption → code execution as root</td></tr>+ <tr><td class="td-key">Introduced</td><td class="td-val">glibc 2.34 (April 2021, commit <code>2ed18c5b534d9e92fc006202a5af0df6b72e7aca</code>, BZ #27471)</td></tr>+ <tr><td class="td-key">Affected version</td><td class="td-val"><code>libc6 2.37-0ubuntu2</code> — confirmed with <code>dpkg -l</code> / <code>ldd --version</code></td></tr>+ <tr><td class="td-key">Patched version</td><td class="td-val"><code>libc6 2.37-0ubuntu2.1</code> — USN-6409-1 (security pocket)</td></tr><tr><td class="td-key">CVSS 3.1</td><td class="td-val"><span class="badge badge-cvss">7.8 HIGH</span></td></tr>- <tr><td class="td-key">Estado explotación</td><td class="td-val"><span class="badge badge-high">CISA KEV · explotación activa</span></td></tr>- <tr><td class="td-key">Exploit público</td><td class="td-val">Sí — PoC funcional (root shell) en <code>github.com/leesh3288/CVE-2023-4911</code>; Qualys TRU obtuvo root completo en Ubuntu 23.04</td></tr>+ <tr><td class="td-key">Exploitation status</td><td class="td-val"><span class="badge badge-high">CISA KEV · active exploitation</span></td></tr>+ <tr><td class="td-key">Public exploit</td><td class="td-val">Yes — working PoC (root shell) at <code>github.com/leesh3288/CVE-2023-4911</code>; Qualys TRU obtained full root on default Ubuntu 23.04 installations</td></tr></table></div>- <!-- ── 3. Evidencia ── -->+ <!-- ── 3. Evidence ── --><div class="section">- <div class="section-title"><span class="section-num">03 /</span> Evidencia de explotabilidad</div>- <p style="color:var(--text-dim);margin-bottom:12px;font-size:13px">Reproductor ejecutado como <code>agent_user</code> (uid=1001) sobre el target:</p>+ <div class="section-title"><span class="section-num">03 /</span> Exploitability Evidence</div>+ <p style="color:var(--text-dim);margin-bottom:12px;font-size:13px">Reproducer executed as <code>agent_user</code> (uid=1001) on the target:</p><div class="terminal"><div class="terminal-bar"><span class="dot dot-red"></span><span class="dot dot-amber"></span><span class="dot dot-green"></span><span>agent_user@ubuntu-server-01 — bash</span></div>- <div class="terminal-body"><span class="comment"># Trigger CVE-2023-4911 — 10 intentos consecutivos</span>+ <div class="terminal-body"><span class="comment"># Trigger CVE-2023-4911 — 10 consecutive attempts</span><span class="cmd">for i in $(seq 1 10); doGLIBC_TUNABLES="glibc.malloc.mxfast=glibc.malloc.mxfast=$(printf 'A%.0s' $(seq 1 1500))" \/usr/bin/su --help 2>&1- echo "intento $i: rc=$?"+ echo "attempt $i: rc=$?"done</span><span class="out-err">Segmentation fault-intento 1: rc=139+attempt 1: rc=139Segmentation fault-intento 2: rc=139+attempt 2: rc=139...Segmentation fault-intento 10: rc=139</span></div>+attempt 10: rc=139</span></div></div>- <p style="color:var(--text-dim);margin:16px 0 10px;font-size:13px">Resultados con controles diferenciales:</p>+ <p style="color:var(--text-dim);margin:16px 0 10px;font-size:13px">Results with differential controls:</p><table class="controls-table">- <thead><tr><th>Prueba</th><th>rc</th><th>Resultado</th></tr></thead>+ <thead><tr><th>Test</th><th>rc</th><th>Result</th></tr></thead><tbody>- <tr><td><code>/usr/bin/su</code> (SUID root) + carga larga, ×10</td><td class="rc-fail">139</td><td class="result-fail">SIGSEGV — Segmentation fault</td></tr>- <tr><td><code>/usr/bin/sudo</code> (SUID root) + carga larga</td><td class="rc-fail">139</td><td class="result-fail">SIGSEGV</td></tr>- <tr><td><code>/usr/bin/mount</code> (SUID root) + carga larga</td><td class="rc-fail">139</td><td class="result-fail">SIGSEGV</td></tr>- <tr><td><code>su --help</code> sin carga (baseline)</td><td class="rc-ok">0</td><td class="result-ok">OK — comportamiento normal</td></tr>- <tr><td><code>/bin/true</code> (no SUID) + misma carga</td><td class="rc-ok">0</td><td class="result-ok">OK — no crash sin SUID</td></tr>- <tr><td><code>mxfast</code> simple 4000 chars (control)</td><td class="rc-ok">0</td><td class="result-ok">OK — no crash sin desbordamiento</td></tr>+ <tr><td><code>/usr/bin/su</code> (SUID root) + long payload, ×10</td><td class="rc-fail">139</td><td class="result-fail">SIGSEGV — Segmentation fault</td></tr>+ <tr><td><code>/usr/bin/sudo</code> (SUID root) + long payload</td><td class="rc-fail">139</td><td class="result-fail">SIGSEGV</td></tr>+ <tr><td><code>/usr/bin/mount</code> (SUID root) + long payload</td><td class="rc-fail">139</td><td class="result-fail">SIGSEGV</td></tr>+ <tr><td><code>su --help</code> without payload (baseline)</td><td class="rc-ok">0</td><td class="result-ok">OK — normal behaviour</td></tr>+ <tr><td><code>/bin/true</code> (non-SUID) + same payload</td><td class="rc-ok">0</td><td class="result-ok">OK — no crash without SUID</td></tr>+ <tr><td><code>mxfast</code> simple 4000 chars (control)</td><td class="rc-ok">0</td><td class="result-ok">OK — no crash without overflow</td></tr></tbody></table>- <p style="color:var(--text-dim);margin:16px 0 8px;font-size:13px">Condiciones habilitantes verificadas en el target:</p>+ <p style="color:var(--text-dim);margin:16px 0 8px;font-size:13px">Enabling conditions verified on the target:</p><div class="condition-pills"><span class="pill active">NoNewPrivs: 0</span>- <span class="pill active">SUID activos: su sudo passwd mount</span>- <span class="pill active">/tmp escribible</span>- <span class="pill active">gcc 12.3.0 instalado</span>- <span class="pill active">libc6 en apt hold</span>- <span class="pill">contenedor Docker</span>- <span class="pill">userspace — no kernel</span>+ <span class="pill active">SUID active: su sudo passwd mount</span>+ <span class="pill active">/tmp writable</span>+ <span class="pill active">gcc 12.3.0 installed</span>+ <span class="pill active">libc6 on apt hold</span>+ <span class="pill">Docker container</span>+ <span class="pill">userspace — not kernel</span></div></div>- <!-- ── 4. Vectores descartados ── -->+ <!-- ── 4. Discarded Vectors ── --><div class="section">- <div class="section-title"><span class="section-num">04 /</span> Vectores evaluados y descartados</div>+ <div class="section-title"><span class="section-num">04 /</span> Evaluated & Discarded Vectors</div><table>- <thead><tr><th>Vector</th><th>Evidencia del descarte</th><th>Estado</th></tr></thead>+ <thead><tr><th>Vector</th><th>Discard evidence</th><th>Status</th></tr></thead><tbody><tr class="discarded-row"><td><code>sudo -l</code></td><td>"Sorry, user agent_user may not run sudo on ubuntu-server-01"</td>- <td><span class="badge badge-info">Descartado</span></td>+ <td><span class="badge badge-info">Discarded</span></td></tr><tr class="discarded-row"><td>File capabilities (<code>getcap -r /</code>)</td>- <td>Sin resultados</td>- <td><span class="badge badge-info">Descartado</span></td>+ <td>No results</td>+ <td><span class="badge badge-info">Discarded</span></td></tr><tr class="discarded-row">- <td>Cron jobs anómalos</td>- <td>Solo <code>e2scrub_all</code> y <code>apt-compat</code> (stock)</td>- <td><span class="badge badge-info">Descartado</span></td>+ <td>Anomalous cron jobs</td>+ <td>Only <code>e2scrub_all</code> and <code>apt-compat</code> (stock)</td>+ <td><span class="badge badge-info">Discarded</span></td></tr><tr class="discarded-row">- <td>Binarios SUID no estándar</td>- <td>Set estándar de Ubuntu (ninguno custom)</td>- <td><span class="badge badge-info">Descartado</span></td>+ <td>Non-standard SUID binaries</td>+ <td>Standard Ubuntu set only (no custom binaries)</td>+ <td><span class="badge badge-info">Discarded</span></td></tr><tr class="discarded-row">- <td>Exploit de kernel</td>- <td>Fuera del modelo de amenaza (contenedor; pista operador: vector userspace)</td>- <td><span class="badge badge-info">Fuera de alcance</span></td>+ <td>Kernel exploit</td>+ <td>Out of threat model scope (container; operator hint: userspace vector)</td>+ <td><span class="badge badge-info">Out of scope</span></td></tr></tbody></table></div>- <!-- ── 5. Remediación ── -->+ <!-- ── 5. Remediation ── --><div class="section">- <div class="section-title"><span class="section-num">05 /</span> Recomendación de remediación</div>+ <div class="section-title"><span class="section-num">05 /</span> Remediation Recommendations</div><ul class="remediation-list"><li>- <span class="priority p1">⚠ Prioridad 1 — Inmediato</span>- <strong>Actualizar libc6 a <code>2.37-0ubuntu2.1</code> o superior</strong> (USN-6409-1).- El paquete está retenido con <code>apt hold</code> — hay que liberarlo primero:+ <span class="priority p1">⚠ Priority 1 — Immediate</span>+ <strong>Update libc6 to <code>2.37-0ubuntu2.1</code> or later</strong> (USN-6409-1).+ The package is pinned with <code>apt hold</code> — it must be unpinned first:<br><br><code>apt-mark unhold libc6 && apt-get update && apt-get upgrade libc6</code><br><br>- ⚠ Ubuntu 23.04 (lunar) está <strong>EOL desde enero 2024</strong>: los repositorios estándar- ya no sirven paquetes. Usar <code>old-releases.ubuntu.com</code> o,- preferentemente, <strong>migrar el contenedor a Ubuntu 24.04 LTS</strong>.+ ⚠ Ubuntu 23.04 (lunar) has been <strong>EOL since January 2024</strong>: standard repositories+ no longer serve packages. Use <code>old-releases.ubuntu.com</code> or, preferably,+ <strong>migrate the container to Ubuntu 24.04 LTS</strong>.</li><li>- <span class="priority p1">⚠ Prioridad 1 — Inmediato tras actualizar</span>- <strong>Reiniciar/reconstruir el contenedor</strong> tras aplicar el parche.- Todos los procesos en ejecución siguen mapeando la <code>libc</code> vulnerable en memoria- hasta que se reinicien.+ <span class="priority p1">⚠ Priority 1 — Immediately after patching</span>+ <strong>Restart/rebuild the container</strong> after applying the patch.+ All running processes continue to map the vulnerable <code>libc</code> in memory+ until restarted.</li><li>- <span class="priority p2">▸ Prioridad 2 — Defensa en profundidad</span>- Auditar periódicamente paquetes en hold: <code>apt-mark showhold</code>. Reducir la- superficie SUID innecesaria. Monitorizar ejecuciones de binarios SUID con- <code>GLIBC_TUNABLES</code> en el entorno (cualquier proceso no-root con esa variable- es sospechoso).+ <span class="priority p2">▸ Priority 2 — Defence in depth</span>+ Periodically audit pinned packages: <code>apt-mark showhold</code>. Reduce unnecessary SUID+ surface. Monitor executions of SUID binaries with <code>GLIBC_TUNABLES</code> in their+ environment — any non-root process carrying that variable should be treated as suspicious.</li><li>- <span class="priority p3">▸ Prioridad 3 — Detección</span>- Comparar <code>ldd --version</code> / <code>dpkg -l libc6</code> contra- <code>2.37-0ubuntu2.1</code> en todos los sistemas Ubuntu 22.04/23.04.- CISA KEV lista CVE-2023-4911 como explotado activamente — priorizar en el siguiente ciclo- de parches.+ <span class="priority p3">▸ Priority 3 — Detection</span>+ Compare <code>ldd --version</code> / <code>dpkg -l libc6</code> against+ <code>2.37-0ubuntu2.1</code> across all Ubuntu 22.04/23.04 systems.+ CISA KEV lists CVE-2023-4911 as actively exploited — prioritise in the next patch cycle.</li></ul></div>- <!-- ── 6. Alcance cumplido ── -->+ <!-- ── 6. Scope ── --><div class="section">- <div class="section-title"><span class="section-num">06 /</span> Alcance cumplido · modo pentest</div>+ <div class="section-title"><span class="section-num">06 /</span> Scope Compliance · mode: pentest</div><ul class="scope-list">- <li><span class="check">✓</span> CVE verificado contra fuentes oficiales (Ubuntu API, NVD, CVE.org, advisory Qualys)</li>- <li><span class="check">✓</span> Crash confirmado del loader privilegiado (10/10 SIGSEGV con controles diferenciales)</li>- <li><span class="check">✓</span> CVSS 3.1 documentado — 7.8 HIGH</li>- <li><span class="check">✓</span> Versiones afectada / parcheada documentadas</li>- <li><span class="check">✓</span> Hallazgos persistidos en el store del engagement (5 findings)</li>- <li><span class="check">✓</span> Bucle OODA cerrado: OBSERVE → ORIENT → DECIDE → ACT → REPORT (14 pasos)</li>+ <li><span class="check">✓</span> CVE verified against official sources (Ubuntu API, NVD, CVE.org, Qualys advisory)</li>+ <li><span class="check">✓</span> Privileged loader crash confirmed (10/10 SIGSEGV with differential controls)</li>+ <li><span class="check">✓</span> CVSS 3.1 documented — 7.8 HIGH</li>+ <li><span class="check">✓</span> Affected and patched versions documented</li>+ <li><span class="check">✓</span> Findings persisted in the engagement store (5 findings)</li>+ <li><span class="check">✓</span> OODA loop closed: OBSERVE → ORIENT → DECIDE → ACT → REPORT (14 steps)</li></ul><div class="scope-note">- <strong>No se compiló el exploit completo ni se obtuvo shell root</strong>, conforme a la- definición de <code>modo: pentest</code>. El camino a root está completamente documentado- (PoC público + condiciones verificadas en el target). Si se desea ejecutar hasta obtener la- shell root, indicarlo explícitamente como <code>modo: redteam</code>.+ <strong>No full exploit was compiled and no root shell was obtained</strong>, in accordance with+ the <code>mode: pentest</code> definition. The path to root is fully documented+ (public PoC + conditions verified on the target). To execute through to root shell, instruct+ the agent explicitly with <code>mode: redteam</code>.</div></div>