Hunt: AD - ADCS expert (microskill)
ADCS exploitation expert of the AD split: certificate-based privilege escalation end-to-end. The mapper (hunt-ad) owns enumeration waves, user lists, and dispatch; this skill owns everything certificate.
Wiki mass: 16 ESC pages + golden-certificate under wiki/techniques/active-directory/ - this skill is the execution shell over that mass.
Shared bypass arsenal (blocked implies research, but walk this first): skills/exploit/hunt/hunt-ad-adcs/references/bypass-arsenal.md - NEW 2025-2026 ESC15/ESC16 classes + the trust-chain bypasses for a hardened CA (Entra External IdP chain, Golden SAML).
Wiki
Hub: [[active-directory-moc]] - Primary page: [[adcs]] - per-ESC pages: active-directory-certificate-esc1 .. esc16 (each page carries its own conditions + exact certipy invocation) Anchors: [[active-directory-golden-certificate]], [[kerberos-attacks]] REFS: [[adcs]], ESC1-16 pages
qmd on hint only: "ADCS ESC certificate template misconfiguration certipy" - per-ESC pages first.
Paths (binding - shared with the mapper run)
# seat side (wsl):
ENG_DIR=/mnt/c/Users/Lenovo/Desktop/VMshare/<eng>/<eng>
RUN=$ENG_DIR/hunt/ad/<domain> # SHARED with hunt-ad
# VM side:
ENG_DIR=/mnt/VMshare/<eng>/<eng>
RUN=$ENG_DIR/hunt/ad/<domain>
Experts write probes/ad-adcs*.jsonl into the mapper's $RUN tree (shared-run precedent). Cert/key material (.pem/.pfx/.ccache) lives ONLY under $RUN/probes/raw/ (leak hygiene; never candidates files).
Transport
Every VM-native command block ships through ONE documented transport:
cd /mnt/c/Users/Lenovo/Documents/ObisidianVaults/ZTorch/ZTorch && bash scripts/vm-ssh.sh '<command>'
- vm-ssh lands as user tz (NOT root); one argument per call; quotes
ship as base64 (
echo <b64> | base64 -d | bash). - Non-login shells do NOT source profile.d: HTTP fetches via /opt/scope-shim/bin/{curl,httpx}; certipy speaks LDAP/HTTPS to the CA directly.
- Out-of-scope dies rc 126 (never work around); UNKNOWN gets an ADVISORY - not approval; self-deny per scope.md.
- certipy req can hang on NETBIOS resolution: ALWAYS pass -dc-ip AND -target-ip . Long sweeps run detached (vm-bg.sh).
Entry gate
All four must pass; each is falsifiable:
Active engagement exists; domain/DC on targets//scope.md. Scope-guard is the deny.
hunt-core loaded (skills/exploit/hunt/hunt-core/SKILL.md).
Mapper verified: $RUN/raw/domain-map.md exists AND a usable principal exists (users.md row or loot.md cred) - ADCS needs SOME authenticated or anon bind - AND the AD-HUNT-DONE pointer ($ENG_DIR/hunt-ad-.md) names adcs in next=. Missing = dispatch hunt-ad first.
Toolchain preflight (fails LOUD):
for t in /usr/bin/jq /usr/bin/python3; do [ -x "$t" ] || { echo "MISSING TOOL: $t"; exit 1; } done command -v certipy >/dev/null || pip3 install --user --break-system-packages -q certipy-ad \ || echo "N/A: certipy absent on VM (install failed)" # manual LDAP template reads remain
Checklist (mandatory)
Copy checklist.md (this skill's dir) to
$ENG_DIR/hunt/ad/<domain>/ad-adcs-checklist.md at run start.
Unchecked at stop = UNCHECKED lane, re-run from the first unchecked
step. SKILL.md and the checklist header agree byte-for-byte on the
copy target.
Nudge
Review-skill re-injection text (pinned; NOTE: completion-gate emits its own generic nudge - both mean: resume at the first unchecked box):
Your work was just reviewed by the review skill and it was found
that you either missed a step of analysis, or didn't generate an
artifact of sufficient proof. Look at the markdown based checklist
and the output for each step, find what you missed, and complete
that step.
Confirmation gate
NOT confirmation: certipy find listing a vulnerable template you never enrolled against; a "vulnerable" flag whose ESC page conditions do not hold on this CA; a cert issued but never used to authenticate.
IS confirmation: a certificate obtained via the ESC exploited AND used to authenticate as the target (certipy auth returns the target's NT hash / a usable TGT; or LDAPS/Kerberos session as the target exercised) - reproduced from your written steps.
Methodology
The 16 ESC pages each carry exact conditions and invocations - this shell sequences them. Rank: ESC1 (misconfigured SAN - the classic), ESC4 (template ACL), ESC8 (web enroll + relay) are the modern high-yield trio; ESC9-16 are certificate-mapping variants gated on UPN/SAN/altSecurityIdentities conditions - CHECK THE PAGE before firing.
Step 1 - Vulnerable-template sweep
certipy find -u <user>@<domain> -p <pass> -dc-ip <dc> \
-vulnerable -stdout [-ldap-scheme ldap]
Runs with ANY cred (the mapper's low-priv principal is enough; ADCS misconfigs leak via default read). Through SOCKS pivots add -ldap-scheme ldap. Every hit -> templates.jsonl row.
Step 2 - CA inventory
CA name(s), web-enrollment presence (/certsrv reachable), enrollment agents, officer rights, target accounts (Administrator, NTLM-hardened ones). @ $RUN/raw/ca-inventory.md.
Steps 3-11 - Per-ESC exploitation
One probes/.jsonl file per class; cert + key archived under probes/raw/; per-ESC wiki page consulted BEFORE firing (conditions first - ESC9-16 especially are mapping-condition-gated):
- ESC1:
certipy req -u <user>@<domain> -p <pass> -ca <ca> -template <vuln> -upn administrator@<domain>-> cert auth as target. - ESC2/3: Any Purpose / enrollment-agent chains (ESC2 any-purpose cert used where EKU-pinning is expected; ESC3 enrollment-agent requests ON BEHALF of the victim).
- ESC4: writable template ACL ->
certipy template -writeto an ESC1 shape, then ESC1. - ESC6/7: SAN inscription on CA / CA-level ACL abuse; ESC7 with backup rights = CA cert + key -> golden certificate (Step 10).
- ESC8: NTLM relay to web enrollment - pair a coercion trigger (printerbug/petitpotam; the wiki network-service-attacks page owns the primitive) with ntlmrelayx to /certsrv, enroll as the relayed victim.
- ESC9-16: certificate-mapping variants - verify the mapping conditions on the CA (how UPN<->SAN<->altSecurityIdentities is resolved) per each ESC page, then the page's exact req/auth pair.
- Shadow credentials: GenericWrite on ANY object ->
certipy shadow auto -account <target>-> LDAPS auth as target. (Feed from the ACL expert's GenericWrite findings.) - Golden certificate: CA cert + private key (ESC7/backup) -> forge for ANY user. RoE-gated: forge for a TEST account, prove auth, stop.
- PKINIT dual-yield: a cert for a NTLM-hardened account yields BOTH
an NT hash and a TGT - when PtH returns STATUS_ACCOUNT_RESTRICTION,
switch to the ccache path (
export KRB5CCNAME=...; impacket-* -k -no-pass). See [[adcs]].
Step 12 - Exit
Append ADCS rows to $RUN/candidates.jsonl ({class: ad-adcs, ...});
update the ADCS section of candidates.md (per-ESC verdicts); append
AD-ADCS-DONE templates=<n> exploited=<n> certs=<n> to
$ENG_DIR/hunt-ad-.md; _requests.md rows (lane = hunt-ad-adcs);
Deadends per clean ESC class naming WHICH conditions failed.
Chaining
- Cert -> TGT -> Kerberos moves: hunt-ad-kerberos.
- GenericWrite feeds (from ACL findings): shadow credentials live here.
- Cert-as-DA -> DCSync close-out: back to the mapper's step 7 region.
Validation
- Proof = the cert AUTHENTICATING as the target (certipy auth hash output / exercised LDAPS or Kerberos session), never the cert file alone.
- Negative control: enrollment against a NON-vulnerable template for the same principal fails.
False Positives
- "Vulnerable" flag with ESC page conditions unmet (checked the page first - this is why).
- Template writable by you but EKU/publisher constraints block the useful cert shape.
- ESC8 relay fired but web enrollment disabled on the CA.
- Cert issued for a user whose SID mapping does not resolve to a privileged context.
Severity
CRITICAL = ESC exploited to a DA-equivalent cert / golden certificate / NTLM-hardened admin via PKINIT. HIGH = cert for privileged non-DA / shadow creds on sensitive object. MEDIUM = misconfig proven, target cert constrained.
N/A escapes
- "N/A: certipy absent on VM (install failed)"
- "N/A: no CA in the domain" (mapper confirmed - whole-run skip) An N/A tick is a claim with a reason, never a silent pass. Identical strings in SKILL.md and checklist.md.
Bitter-pill variant
Freehand run: skip checklist + numbered steps, one freehand pass over the CA, output ONLY $RUN/analysis/bitter-pill.md (tried/found/ dismissed, every claim carries evidence). The cert-authenticated gate survives freehand. Declare with a _requests.md row (lane = hunt-ad-adcs-bitterpill).
Depth walk (mandatory)
Walk EVERY applicable row of the payload arsenal page cited above.
Record each row in $RUN/coverage-depth.md as a table
| row | state | evidence | where state is probed, blocked, or
N/A, and evidence is a probe-pair path or a Deadends.md line. A bare
probed with no evidence reference fails the completion gate
(NEEDS-REVIEW). N/A rows cite why (surface shape forbids it).
Blocked implies research: before any Deadend is claimable on a blocked
row, (a) walk the arsenal page's bypass-ladder rows for that control
class, and (b) run one wiki query AND one mcp__wigolo__search for the
specific tech-plus-control bypass. Results, or a genuine null, land in
coverage-depth.md. "No research tried" is not a claimable state.
Exit artifact + handoff
Appends into the shared mapper $RUN: probes/ad-adcs*.{jsonl,md},
raw/ca-inventory.md, candidates.jsonl/md (ADCS rows), pointer line
AD-ADCS-DONE. Cert material ONLY under probes/raw/. Consumer: triage.
Findings leave ONLY through G1. Distill: python3 scripts/wiki-stage.py --kind technique --slug <slug> --target-page techniques/active-directory/adcs.md.
Stop conditions
All ESC steps terminal; forged certs only for TEST accounts (auth proven, no persistence); nothing written into skill dirs; no unchecked boxes above REVIEW; hunt-core stop conditions fire unchanged.
Attack surface
APPROACH: Fire the unauth-enum wave at once (null/guest shares + RID-brute, anon LDAP users + description fields, AS-REP roast, person-object SPN Kerberoast, lockout policy, certipy find -vulnerable), then walk enum -> roast -> ACL/ADCS ESC1-16 -> delegation -> DCSync -> lateral, reusing captured creds inside the lockout gate.
AVOID: A captured-but-uncracked hash, a dumped ticket, a theoretical BloodHound/ESC edge never executed, or an unvalidated spray hit is NOT confirmation - creds must authenticate to the DC ([+]/Pwn3d!) or the ticket/ACL abuse must be exercised end to end.
Cross-routing
- Load hunt-windows when you land on a member host for local privesc before DA.
- Load hunt-m365 when the tenant is Entra/hybrid (AD Connect bridge both ways).
- Load hunt-auth for AD FS / cert-based web auth flows.
- Not this skill: plain local privesc with no domain in play (hunt-windows).