22-purple-team
Collaborative purple-team operations — threat-informed adversary emulation
Install / Use
npx skills add Masriyan/Claude-Code-CyberSecurity-Skill --skill 22-purple-teamInstalls into whichever agent you are using.
SKILL.md
Installable skill definition
Quality Score
Category
SecuritySupported Platforms
Our assessment of 22-purple-team
22-purple-team scores 88/100 on our quality scale, 645th of 1,097 Security skills we index.
Its SKILL.md is 12 KB long, well organised into 27 sections with 5 code examples: a thorough specification that gives an agent plenty to work with.
It has 441 GitHub stars, a meaningful sign that others use it.
Maintenance, license and trust
- The repository was last updated about 2 months ago, so 22-purple-team is actively maintained.
- Our last check on 2026-11-01 found the source still online.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-10-07. Automated pattern scan on 2026-10-07. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
22-purple-team compared with similar skills
All 4 of these similar skills score higher than 22-purple-team; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| 22-purple-team (this skill)by Masriyan | 88 | 441 | 2mo ago | SKILL.md |
| Agent-Reachby Panniantong | 100 | 113.0k | 10d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 80.1k | today | CLAUDE.md |
| Scraplingby D4Vinci | 100 | 104.2k | 1d ago | MCP Server |
| crawl4aiby unclecode | 100 | 87.5k | today | MCP Server |
Frequently asked questions
- How do I install 22-purple-team?
- Run
npx skills add Masriyan/Claude-Code-CyberSecurity-Skill --skill 22-purple-team. The install tabs above show the steps for each supported agent. - Which AI agents does 22-purple-team work with?
- It is written for Claude Code, as a SKILL.md file. Other agents that read the same format can often use it too.
- Is 22-purple-team safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is 22-purple-team still maintained?
- The repository was last updated about 2 months ago, so 22-purple-team is actively maintained.
Skill content
View source on GitHubname: Purple Team & Adversary Emulation description: Collaborative purple-team operations — threat-informed adversary emulation planning (ATT&CK, CTID, Atomic Red Team, CALDERA), the detect-tune-validate loop, detection coverage measurement (DeTT&CT/Navigator), safe execution and deconfliction, and MTTD/coverage reporting version: 3.1.0 author: Masriyan tags: [cybersecurity, purple-team, adversary-emulation, attack, atomic-red-team, caldera, detection-engineering, detection-coverage, mttd, breach-attack-simulation]
Purple Team & Adversary Emulation
Purpose
Enable Claude to plan and run purple team engagements: red and blue working the same ATT&CK techniques together so that every emulated behavior produces a measured answer to "did we see it, and could we have stopped it?" The deliverable is not a compromise — it is a validated, quantified improvement in detection and prevention coverage.
This skill sits between Skill 14 (Red Team — how to execute the offensive TTPs) and Skills 15/12/11 (Blue Team, SIEM/Sigma, SOC — how to detect and respond). It supplies the collaborative loop and the coverage measurement that turn one-off findings into durable detections.
⚠️ Authorization Gate — Read First
Adversary emulation executes real attack techniques against real systems. Before providing operational assistance (test execution, payloads, C2, live-fire steps), confirm:
[ ] Written authorization naming the systems/environment in scope exists and is current
[ ] Scope, allowed techniques, and explicit exclusions (fragile prod, safety systems, OT) are defined
[ ] A deconfliction channel and point of contact are agreed with the SOC/blue team
[ ] A stop/abort procedure and rollback plan are documented and understood
[ ] Destructive or availability-impacting techniques are excluded unless separately authorized
[ ] Test data/accounts are used; no real user data is exfiltrated
If authorization cannot be confirmed, restrict assistance to planning, coverage analysis, detection engineering, and tabletop emulation — which need no live execution — and say so explicitly. Never provide live-fire steps against systems the user cannot show authorization for. For the offensive tradecraft itself, defer to Skill 14's Rules of Engagement.
Activation Triggers
This skill activates when the user asks about:
- Planning a purple team engagement, adversary emulation plan, or breach-and-attack-simulation exercise
- Choosing and emulating a specific threat actor's TTPs (from CTI / ATT&CK groups)
- Atomic Red Team, MITRE CALDERA, CTID Adversary Emulation Library, or MITRE Engenuity ATT&CK Evaluations
- The detect → tune → re-test validation loop for a technique
- Measuring or visualizing detection coverage (ATT&CK Navigator layer, DeTT&CT, heatmap)
- Detection maturity per technique (none / telemetry / detection / prevention)
- Deconfliction between red and blue, or safe execution of attack techniques
- Purple-team metrics — MTTD/MTTR, detection rate, per-technique time-to-detect, gap backlog
Prerequisites
python3 --version # 3.10+; standard library only for the core script
pip install pyyaml # optional — only to load emulation plans written in YAML
Optional enhanced tooling:
- Atomic Red Team — library of small, ATT&CK-mapped tests
- MITRE CALDERA — automated adversary emulation platform
- CTID Adversary Emulation Library — full actor emulation plans
- DeTT&CT — data-source and detection coverage scoring
- ATT&CK Navigator — coverage heatmaps
- VECTR — purple-team assessment tracking
Core Capabilities
1. Engagement Model Selection
Pick the collaboration model that fits the maturity and goal:
- Tabletop emulation — walk the actor's kill chain on paper; verify each step would generate telemetry and should fire a detection. No execution; safe anywhere, ideal first pass.
- Micro-emulation — a handful of atomic tests for a specific technique or a newly deployed detection. Fast feedback, small blast radius.
- Full-campaign emulation — end-to-end emulation of a named actor across the kill chain, run collaboratively with the SOC watching in real time. Highest fidelity, highest coordination cost.
- Continuous / automated (BAS) — scheduled automated tests to catch detection regressions over time.
Default to the least invasive model that answers the question, and escalate only with cause and authorization.
2. Threat-Informed Emulation Planning
When the user asks to plan an emulation:
- Choose the adversary from intelligence, not convenience. Pull the actor/campaign relevant to the org's sector and crown jewels (hand-off from Skill 21 CTI / Skill 06). Prefer an actor the org actually faces.
- Build the technique list by walking the actor's known TTPs across ATT&CK tactics (Initial Access → Impact). For each, capture: technique ID, tactic, a concrete procedure, the expected data source/telemetry, and the expected detection.
- Order by kill chain and set objectives — what "success" means for each step (e.g., "Kerberoast a service account" → objective: SOC alerts within N minutes).
- Map to a test source — an Atomic Red Team test, a CALDERA ability, or a documented manual procedure — so execution is repeatable.
- Record the plan as structured data (
scripts/detection_validator.pyreads a JSON/YAML plan) so results are trackable across re-tests.
3. The Detect–Tune–Validate Loop
For every technique, run the collaborative loop and record the outcome:
Execute technique (red, announced to blue)
↓
Confirm telemetry exists ──no──▶ GAP: missing data source → Skill 15 (onboard log source)
↓ yes
Confirm a detection fires ──no──▶ GAP: no detection → Skill 12 (write Sigma) / Skill 15
↓ yes
Was it prevented? ──no──▶ note: detect-only → hardening backlog → Skill 15
↓
Measure time-to-detect, tune false positives, re-test to confirm the fix
Classify each technique's outcome on this ladder: none (no telemetry) → telemetry (data exists, no alert) → detection (alert fires) → prevention (blocked). The gap between telemetry and detection is where most purple-team value is realized.
4. Coverage Measurement & Visualization
- Score every in-scope technique on the none/telemetry/detection/prevention ladder and compute coverage per tactic and overall.
- Produce an ATT&CK Navigator layer (technique IDs colored by coverage) so the heatmap is shareable and diff-able across engagements.
- Track drift — compare this engagement's layer to the last one; a technique that regressed from detection→telemetry is a finding.
- Avoid the coverage-theater trap — a technique "covered" by one brittle, high-false-positive rule is not covered. Weight robustness and note single-signature dependencies.
- Use
scripts/detection_validator.pyto turn a results plan into a per-tactic coverage matrix, a Navigator layer, MTTD statistics, and a prioritized gap list.
5. Safe Execution & Deconfliction
- Announce every action to the blue team in the agreed channel with a timestamp, so a real intrusion during the exercise is not mistaken for the test (and vice versa).
- Bound the blast radius — test accounts, test data, non-production or clearly-scoped hosts; never destructive/availability-impacting techniques without separate sign-off (see Skill 18 for OT/ICS where impact can be physical).
- Have an abort — a single agreed signal that stops all offensive activity, plus a rollback for any change made.
- Log everything — command, time, operator, target, expected vs. observed — so results are attributable and repeatable.
6. Metrics & Reporting
Report outcomes, not activity:
- Detection rate — % of executed techniques that fired a detection; prevention rate — % blocked.
- MTTD / MTTR — mean time to detect / respond, per technique and overall; call out the slowest.
- Coverage delta — improvement vs. the previous engagement.
- Remediation backlog — each gap as an actionable item (missing log source, new detection to write, hardening change) with an owner and priority, handed to Skills 15/12/11.
Output Template
# Purple Team Engagement Report
**Engagement:** [Name] | **Date:** [YYYY-MM-DD]
**Emulated adversary:** [Actor / campaign] | **Model:** [Tabletop / Micro / Full-campaign / BAS]
**Authorization:** [reference] | **Deconfliction POC:** [name/channel]
---
## Executive Summary
[2–3 sentences: techniques run, detection rate, prevention rate, top gap, coverage delta.]
## Coverage Scorecard
| Metric | Result |
|--------|--------|
| Techniques executed | N |
| Detected | N (xx%) |
| Prevented | N (xx%) |
| Mean time-to-detect | mm:ss |
| Coverage delta vs. prior | +/- x techniques |
## Per-Technique Results
| Tactic | Technique (ID) | Executed | Data source | Outcome | TTD | Detection / Gap |
|--------|----------------|----------|-------------|---------|-----|-----------------|
| Credential Access | Kerberoasting (T1558.003) | ✅ | Security 4769 | detection | 04:12 | Sigma: rc4 TGS request |
| Defense Evasion | Clear Windows Event Logs (T1070.001) | ✅ | Security 1102 | telemetry | — | GAP: no alert on 1102 |
## Coverage Heatmap
[ATT&CK Navigator layer attached: `navigator_layer.json`]
## Prioritized Remediation Backlog
| Priority | Gap | Action | Owner | Handoff |
|----------|-----|--------|-------|---------|
| High | No alert on log clearing | Write Sigma for EID 1102 | Detection Eng | → Skill 12 |
| Med | RDP lateral movement detect-only | Add prevention control | Blue Team | → Skill 15 |
Script Reference
detection_validator.py
# Score an emulation results plan: coverage matrix, MTTD, gaps
python scripts/detection_validator.py --plan results.json
# Also emit an ATT&CK Navigator layer and a JSON report
python scripts/detection_validator.py --plan results.json \
--navigator navigator_layer.json --output report.json
# Compare against a previous engagement to compute coverage drift
python scripts/detection_validator.py --plan results.json --baseline prior.json
# Self-contained demo with a built-in sample emulation plan
python scripts/detection_validator.py --demo
A plan is a JSON (or YAML with pyyaml) list of test entries — technique_id, technique, tactic, executed, data_source, outcome (none/telemetry/detection/prevention), and optional time_to_detect_seconds. The script computes per-tactic and overall coverage, detection/prevention rates, MTTD, a prioritized gap list, and an ATT&CK Navigator layer. Scoring reflects only what the plan records — it validates your results, it does not execute anything.
Skill Integration
| Condition | Next Skill | |-----------|------------| | Which adversary/TTPs to emulate | ← Skill 21 (CTI) and Skill 06 (Threat Hunting) | | Executing the offensive techniques | → Skill 14 (Red Team Operations) — under its RoE | | A gap needs a detection written | → Skill 12 (Log Analysis / Sigma) | | A gap needs a hardening/prevention control | → Skill 15 (Blue Team Defense) | | Validating SOC alerting/triage in the loop | → Skill 11 (CSOC Operations) | | Emulation touches OT/ICS environments | → Skill 18 (OT/ICS) — safety-first, non-disruptive only | | A real incident is discovered mid-exercise | → Skill 07 (Incident Response) — declare and deconflict |
References
- MITRE ATT&CK · ATT&CK Navigator
- [Center for Threat-Informed Def
Truncated for display — read the full file on GitHub.
Related Skills
Agent-Reach
113.0kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
headroom
80.1kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
Scrapling
104.2k🕷️ The BEST Web Scraping framework that handles everything from a single request to a full-scale crawl! Don't be shy, join here: https://discord.gg/EMgGbDceNQ and follow here for daily tips and tricks: https://x.com/Scrapling_dev
crawl4ai
87.5kOfficial Crawl4AI Cloud: a hosted web scraping, extraction, and search API for AI agents. Free API key, 1,000 credits a month, no card. Fully open source: this repo is the Apache-2.0 crawler that powers it, and you can self-host it.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
