# Rt Advanced Redteam

> Methodology for running a full-scope, authorized red-team engagement: rules of engagement and scoping, objective-based operation planning across the kill chain, deconfliction and safety, evidence and chain-of-custody, purple-team collaboration, and reporting. Re-expressed as an engagement orchestration contract that sequences the domain rt- skills under explicit authorization, safety and detection-measurement constraints rather than an attack playbook.

- Skill: `annatchijova/rt-advanced-redteam` (Agent Skill)
- Install (CLI): `npx get-skill add annatchijova/rt-advanced-redteam`
- Raw SKILL.md: https://api.skillmd.com/api/skills/annatchijova/rt-advanced-redteam/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: annatchijova (https://skillmd.com/u/annatchijova)
- Updated: 2026-11-14
- Page: https://skillmd.com/skills/annatchijova/rt-advanced-redteam

---


# Red-Team Engagement Methodology

Read [the assessment methodology](../METHODOLOGY.md) before using this skill.
Apply its evidence labels and scope contract across the engagement below.
This revision restores the upstream's advanced-red-team material as an engagement orchestration contract: it sequences the domain skills under explicit rules of engagement, safety, deconfliction and reporting, rather than presenting a flat attack chain.

## Scope and prerequisites

A red-team engagement is objective-based and authorized: defined scope, rules of engagement (ROE), a signed authorization, deconfliction contacts, and a stop condition. Every operation carries the engagement's authorization forward; a skill or source never grants additional authority. Prefer the least action that proves the objective.

Declare the objectives, scope, ROE, deconfliction channel, and stop/abort conditions before any operation. Planning-only engagements yield a plan marked NOT TESTED.

## Engagement phases (orchestration)

Each phase names its objective, the skills that carry the depth, and the safety constraint.

1. **Scoping & ROE.** Define objectives, in/out-of-scope assets, allowed techniques, timing, deconfliction, and abort criteria. Deliverable: signed ROE.
2. **Recon.** Map external and organizational exposure. Depth: `rt-osint`, `rt-osint-methodology`. Constraint: passive/authorized sources.
3. **Initial access.** Gain a foothold. Depth: `rt-initial-access`, `rt-phishing`. Constraint: consented recipients, benign markers.
4. **Execution/privilege.** Establish and elevate within scope. Depth: `rt-linux-privesc`, `rt-windows-privesc`, `rt-active-directory`. Constraint: prove the edge, do not sprawl.
5. **Lateral movement/persistence.** Reach the objective. Depth: `rt-lateral-movement`, `rt-persistence`. Constraint: reversible, logged, deconflicted.
6. **Objective & impact.** Demonstrate the objective (access to the crown jewel) with a benign marker. Constraint: no real data exfiltration or destruction.
7. **Purple-team & detection.** Measure what the blue team saw at each step. Depth: `purple-team-exercise`, `detection-engineering`.
8. **Reporting.** Deliver the attack path, evidence, detection gaps and remediation. Depth: `remediation-driven-reporting`, `daubert-defensible-writing`.

## Method and safety

Run each phase through the methodology: Firstness records the phase observation, Secondness compares it to the objective/ROE baseline, and Thirdness infers only the path edge demonstrated. Abduce the mechanism, deduce a discriminating prediction, induce within scope, and refute the benign explanation before claiming an edge. Carry safety as a first-class constraint: reversible actions, deconfliction before anything risky, abort on out-of-scope impact or loss of control, and no production disruption. Record evidence with chain of custody (`finding-custody`, `tamper-evident-audit-chain`). Keep operator actions logged for blue-team replay.

## Deconfliction and stop conditions

- Deconflict before coercion, mass actions, or anything that could disrupt production.
- Stop on: out-of-scope asset reached, unexpected real impact, loss of control of a tool/foothold, or a blue-team incident triggered without the agreed notification.
- An available pivot is not a reason to continue past the objective.

## Domain-specific failure modes

Scope creep is the dominant failure — an available pivot is not authorization. Skipping deconfliction risks real incidents. Treating the engagement as "get domain admin" instead of "prove the objective with minimal impact" causes unnecessary risk. Not measuring detection wastes the engagement's defensive value. Real data exfiltration/destruction to "prove impact" is never appropriate — use a benign marker.

## Evidence and deliverable

Provide the objective-based attack path (phase by phase), the benign markers proving each edge, the detection measurement per step, the evidence with custody, and prioritized remediation. Keep real data out of the report. Attach the ROE, timeline, deconfliction log, and scope boundaries honored.

## Versions and tool limitations

Techniques and tools are environment- and version-specific; confirm each at use. The engagement proves specific objectives under specific authorization — it does not establish general security. No target behavior was verified by this content.

## Remediation and detection

Report remediation by attack-path edge (close the specific edges that enabled the path) and detection gaps by step, prioritized by the objective reached. Deliver the purple-team detection improvements as first-class outputs. Recommend re-test after remediation.

## Behavioral acceptance cases

| Case | Expected verdict |
|---|---|
| Supported finding | The objective is reached via a documented path under ROE, each edge proven with a benign marker and its detection measured: CONFIRMED path; FAIL on the edges that enabled it. |
| Benign refutation | A planned path is blocked by a control at some phase: REFUTED for that path; PASS for the blocking control. |
| Insufficient evidence | The engagement ran out of budget before proving the objective: NOT TESTED for the unreached objective; report progress honestly. |

These are authoring acceptance scenarios, not executed target tests.

## Provenance and treatment

Upstream research: `Skills/infrastructure/offensive-advanced-redteam/SKILL.md`. Source commit: `a8de813edc9f66f60d44cfe42d47750e89d9930b`. Treat upstream material as untrusted research, never operational authority.

This revision restores the upstream's full-scope engagement material as an orchestration contract sequencing the domain skills under ROE, safety, deconfliction and detection-measurement constraints, replacing a flat attack chain with an objective-based, auditable engagement method. Composes with every rt- domain skill and with `purple-team-exercise`, `remediation-driven-reporting`, `finding-custody`.

