Kill-chain orchestration
When it applies
An authorized pentest (never bug bounty — persistence and lateral movement exceed program
scopes) whose objective spans stages: "external to domain admin", "full internal pentest",
"I have a webshell — what's the route", a red-team exercise. tradecraft-attack-scenarios
turns the objective into an ordered plan and tradecraft-attack-path-mapping models the
graph; this skill is the execution layer — phase gates, per-phase playbooks, and the
discipline that keeps a multi-day operation coherent. Single-stage tasks skip it and go
straight to their domain skill (web-*, ad-*, cloud-*).
Why it works
Multi-stage engagements fail by drift, not by missing techniques. Re-planning at each phase
gate — what do I hold, what does it unlock, what's the cheapest next edge — reaches the
objective with least action and noise. The kill chain is a routing spine; each phase delegates
to the specialist skills that own the technique, and a blocked phase reroutes instead of
grinding.
Method
Full playbooks (web→DC, phishing→internal, proximity, cloud, AD CS) and the
state→next-step decision matrix: cheatsheet.md.
- Gate: scope first. Load
tradecraft-scope-roe; confirm the signed scope in
engagements/<target>/scope.txt. Persistence, EDR evasion, and log tampering need explicit
written authorization — assumed only under a red-team RoE, never by default.
- Recon (
recon-*): subdomains → live hosts (httpx) → ports (naabu/nmap) → tech
fingerprint (nuclei -tags tech). Prioritize test/dev/staging environments and newly
deployed systems; mine leaks (recon-github-code-leaks) for cloud keys, connection
strings, JWT secrets. Gate: asset list + ranked candidate leads in plan.md.
- Initial access — highest-signal vector first: known-vuln components
(
nuclei -severity critical,high), the web classes (web-sqli, web-ssti,
web-file-upload, web-ssrf → cloud-imds-ssrf for cloud creds), perimeter appliances
(network-appliance-attacks), exposed services (network-service-attacks), or
phishing/physical when the RoE includes them (social-eng-* — load
social-eng-methodology first). Gate: a foothold with proven command execution.
- Privilege escalation (
privesc-enumeration before any exploit): Windows token abuse
(privesc-windows-tokens), sudo/SUID (privesc-linux-gtfobins), cloud IAM
(cloud-iam-privesc). Gate: admin/SYSTEM/root, or a more powerful identity than before.
- Lateral movement — loot credentials (
privesc-enumeration, mimikatz/secretsdump), then
move: ad-* (Kerberoasting, delegation, DACL, ADCS), network-ntlm-relay, WinRM/SSH, and
network-pivoting-tunneling for segmented networks. BloodHound drives the path
(tradecraft-attack-path-mapping). Gate: every hop is a validated edge, not an
assumption.
- Objective — reach what the engagement exists to prove: domain admin, the data store,
the crown-jewel system. Prove with the least data possible (anonymize any proof data).
Implants/persistence only if the RoE authorizes them and a cleanup date is agreed.
- Close out — remove every artifact you introduced (accounts, uploads, scheduled tasks,
beacons, dropped tools, added SSH keys), restore what you changed, reconcile against your
activity log. Then
reporting-evidence-review → reporting-pentest-report.
Re-plan at every gate: a blocked phase routes back through tradecraft-pivot-decisions to the
next-best edge.
Gotchas
- Don't blast the whole chain with automation against a domain — an unscoped scanner run
is how engagements end careers. Work the asset list, phase by phase.
- One failed technique ≠ blocked phase — work the variation matrix (House rules) and the
alternate edges before declaring a gate unpassable.
- High-risk actions (DCSync, golden ticket, log tampering) need explicit RoE coverage —
if it isn't written down, don't do it; propose it to the client instead.
- Honeypots: an oddly open share or a too-tempting credential is a trap — verify before
you use it.
- Never touch availability or real user data — no DoS, no mass exfiltration.
- Classic failure modes to avoid: mimikatz dumps left on disk, C2 on a threat-intel-burned
domain, phishing that trips DLP because the gateway was never tested, lateral movement that
springs a honeypot.
Verify success
Every phase gate has a written pass/fail in plan.md; the path from first foothold to
objective is reproducible edge by edge; all introduced artifacts are removed and accounted
for; the evidence chain survives reporting-evidence-review.
References
MITRE ATT&CK tactics; PTES; tradecraft-attack-scenarios, tradecraft-attack-path-mapping,
tradecraft-complex-engagements.
Portions adapted from reverse-skill by zhaoxuya520, MIT License.
1---2name: tradecraft-attack-chain3description: >4 Orchestrate a full multi-stage kill chain — recon → initial access → privilege escalation →5 lateral movement → objective — with phase gates, per-phase playbooks, and operating6 discipline. Load on pentest tasks that span stages: "get from external to domain admin",7 "full internal pentest", "I have a webshell, route me to the objective", red-team exercise8 planning. Single-stage tasks go straight to their domain skill; bug bounty never uses this.9domain: tradecraft10type: methodology11stability: learning12modes: [pentest]13severity: high14mitre: [T1190, T1078, T1021, T1558]15tools: []16schema_version: 117metadata:18 source: https://skillmd.com/skills/noorqureshi/tradecraft-attack-chain19---2021# Kill-chain orchestration2223## When it applies24An authorized **pentest** (never bug bounty — persistence and lateral movement exceed program25scopes) whose objective spans stages: "external to domain admin", "full internal pentest",26"I have a webshell — what's the route", a red-team exercise. `tradecraft-attack-scenarios`27turns the objective into an ordered plan and `tradecraft-attack-path-mapping` models the28graph; this skill is the execution layer — phase gates, per-phase playbooks, and the29discipline that keeps a multi-day operation coherent. Single-stage tasks skip it and go30straight to their domain skill (`web-*`, `ad-*`, `cloud-*`).3132## Why it works33Multi-stage engagements fail by drift, not by missing techniques. Re-planning at each phase34gate — what do I hold, what does it unlock, what's the cheapest next edge — reaches the35objective with least action and noise. The kill chain is a routing spine; each phase delegates36to the specialist skills that own the technique, and a blocked phase reroutes instead of37grinding.3839## Method40> Full playbooks (web→DC, phishing→internal, proximity, cloud, AD CS) and the41> state→next-step decision matrix: [`cheatsheet.md`](cheatsheet.md).42430. **Gate: scope first.** Load `tradecraft-scope-roe`; confirm the signed scope in44 `engagements/<target>/scope.txt`. Persistence, EDR evasion, and log tampering need explicit45 written authorization — assumed only under a red-team RoE, never by default.461. **Recon** (`recon-*`): subdomains → live hosts (`httpx`) → ports (`naabu`/`nmap`) → tech47 fingerprint (`nuclei -tags tech`). Prioritize test/dev/staging environments and newly48 deployed systems; mine leaks (`recon-github-code-leaks`) for cloud keys, connection49 strings, JWT secrets. **Gate: asset list + ranked candidate leads in `plan.md`.**502. **Initial access** — highest-signal vector first: known-vuln components51 (`nuclei -severity critical,high`), the web classes (`web-sqli`, `web-ssti`,52 `web-file-upload`, `web-ssrf` → `cloud-imds-ssrf` for cloud creds), perimeter appliances53 (`network-appliance-attacks`), exposed services (`network-service-attacks`), or54 phishing/physical when the RoE includes them (`social-eng-*` — load55 `social-eng-methodology` first). **Gate: a foothold with proven command execution.**563. **Privilege escalation** (`privesc-enumeration` before any exploit): Windows token abuse57 (`privesc-windows-tokens`), sudo/SUID (`privesc-linux-gtfobins`), cloud IAM58 (`cloud-iam-privesc`). **Gate: admin/SYSTEM/root, or a more powerful identity than before.**594. **Lateral movement** — loot credentials (`privesc-enumeration`, mimikatz/secretsdump), then60 move: `ad-*` (Kerberoasting, delegation, DACL, ADCS), `network-ntlm-relay`, WinRM/SSH, and61 `network-pivoting-tunneling` for segmented networks. BloodHound drives the path62 (`tradecraft-attack-path-mapping`). **Gate: every hop is a validated edge, not an63 assumption.**645. **Objective** — reach what the engagement exists to prove: domain admin, the data store,65 the crown-jewel system. Prove with the least data possible (anonymize any proof data).66 Implants/persistence only if the RoE authorizes them and a cleanup date is agreed.676. **Close out** — remove every artifact you introduced (accounts, uploads, scheduled tasks,68 beacons, dropped tools, added SSH keys), restore what you changed, reconcile against your69 activity log. Then `reporting-evidence-review` → `reporting-pentest-report`.7071Re-plan at every gate: a blocked phase routes back through `tradecraft-pivot-decisions` to the72next-best edge.7374## Gotchas75- **Don't blast the whole chain with automation against a domain** — an unscoped scanner run76 is how engagements end careers. Work the asset list, phase by phase.77- **One failed technique ≠ blocked phase** — work the variation matrix (House rules) and the78 alternate edges before declaring a gate unpassable.79- **High-risk actions (DCSync, golden ticket, log tampering) need explicit RoE coverage** —80 if it isn't written down, don't do it; propose it to the client instead.81- **Honeypots**: an oddly open share or a too-tempting credential is a trap — verify before82 you use it.83- **Never touch availability or real user data** — no DoS, no mass exfiltration.84- Classic failure modes to avoid: mimikatz dumps left on disk, C2 on a threat-intel-burned85 domain, phishing that trips DLP because the gateway was never tested, lateral movement that86 springs a honeypot.8788## Verify success89Every phase gate has a written pass/fail in `plan.md`; the path from first foothold to90objective is reproducible edge by edge; all introduced artifacts are removed and accounted91for; the evidence chain survives `reporting-evidence-review`.9293## References94MITRE ATT&CK tactics; PTES; `tradecraft-attack-scenarios`, `tradecraft-attack-path-mapping`,95`tradecraft-complex-engagements`.9697---98_Portions adapted from [reverse-skill](https://github.com/zhaoxuya520/reverse-skill) by zhaoxuya520, MIT License._