Skills coordinator
📦

coordinator

Content revision r1 Critical ⚙️ External commands🌐 Network access

Coordinate CTF Penetration Tests

Authorized security labs often require tracking across recon, exploitation, and privilege escalation. This skill coordinates agent delegation and state-driven decisions for controlled CTF-style assessments.

Supports: Claude Codex Code(CC)
⚠️ 38 Poor

Install with my Agent

Copy this request to your Agent. It includes the canonical Skill page and manifest.

Agent request
Review the Skillstore skill "coordinator" from https://skillstore.io/skills/charleskozel-coordinator.md and its manifest at https://skillstore.io/api/skills/charleskozel-coordinator/manifest. Verify the artifact. Do not auto-install. Inspect the skill and report your findings, then wait for an operator or manual installation decision.

Your Agent should still show its plan and request any confirmation required by the security policy.

Agent-readable resources

Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.

Markdown detail GET /skills/charleskozel-coordinator.md Signed manifest GET /api/skills/charleskozel-coordinator/manifest Signed lockfile GET /api/skills/charleskozel-coordinator/lockfile

Test it

Using "coordinator". Begin an authorized CTF run for a lab target.

Expected outcome:

The skill identifies reconnaissance as the current phase, assigns an Explore task, and waits for service findings before choosing a path.

Using "coordinator". An agent found web directories but no shell.

Expected outcome:

The skill reviews tried vectors, prioritizes a focused follow-up task, and keeps the work inside the approved lab scope.

Using "coordinator". The lab reports user access but no root flag.

Expected outcome:

The skill moves to privilege-escalation coordination and requests evidence-driven checks before marking the exercise complete.

Security Audit

Critical
v8 • 7/6/2026 Open versioned report

Several static code-fence detections were false positives, but the confirmed findings show explicit network reconnaissance, exploitation, and privilege-escalation orchestration. Semantic review found critical autonomous offensive behavior, non-termination prompt injection, and an Active Directory credential-abuse playbook.

2
Files scanned
760
Lines analyzed
7
Review items
0
False positives ignored

Confirmed security concerns (5)

Critical
SKILL.md:3SKILL.md:21-41SKILL.md:114-115
Autonomous Offensive Penetration Testing Workflow
The skill automatically activates on a target IP and coordinates reconnaissance, exploitation, privilege escalation, and flag capture through delegated agents.
The description and task prompts directly name reconnaissance, exploitation, privilege escalation, and repeated delegation until flags are captured. The offensive intent is explicit.
Critical
AD_ATTACK_SUPPLEMENT.md:3-5AD_ATTACK_SUPPLEMENT.md:22-33AD_ATTACK_SUPPLEMENT.md:528-540
Prompt Injection Attempt Detected
The supplement states "ABSOLUTE RULE" and says the agent cannot stop except after both flags are captured. It also requires indefinite continuation until flags are found.
The file uses mandatory override language that prevents stopping, asking for help, or ending when stuck. This attempts to constrain agent behavior beyond normal user and safety controls.
Critical
AD_ATTACK_SUPPLEMENT.md:75-108AD_ATTACK_SUPPLEMENT.md:360-388AD_ATTACK_SUPPLEMENT.md:392-404AD_ATTACK_SUPPLEMENT.md:431-492
Active Directory Credential Abuse Playbook
The supplement gives concrete techniques for empty-password authentication, LDAP password modification, Kerberos ticket use, SMB execution, LAPS password access, and password-variation attacks.
The cited sections contain named tools, authentication bypass attempts, password-change flows, and sensitive attribute access. These are concrete credential-abuse procedures.
High
AD_ATTACK_SUPPLEMENT.md:402-403
Network reconnaissance
- If "Everyone" or "Anonymous" is member, extensive read access
The supplement instructs abuse of Everyone or Anonymous membership to gain extensive read access and possibly read LAPS passwords. That is credential-access guidance, not benign reconnaissance.
High
SKILL.md:15
Network scanning tools
- Running nmap scans
Nmap scanning is explicitly listed as work the coordinator should delegate to specialized agents. The surrounding section confirms the skill is designed to orchestrate network reconnaissance.
Capability review items (7)

These are real local capabilities that may be expected for this skill, so they require review but are not counted as confirmed malicious behavior.

High
SKILL.md:21-45
Ruby/shell backtick execution
```python
The block contains actionable agent prompts to scan all ports, exploit Apache, and escalate privileges. Although written as Markdown, it is intended to drive external offensive actions through subagents.
High
SKILL.md:119-156
Ruby/shell backtick execution
```
The example flow starts a target assessment, delegates scanning, enumerates web paths, tests uploads, and then escalates to root. This is actionable offensive automation even though it is presented as an example.
High
SKILL.md:170-177
Ruby/shell backtick execution
```python
The Task prompt tells a subagent to try alternative attack vectors and research service-version CVEs after failure. That can continue offensive activity through delegated external tooling.
High
SKILL.md:17
sudo privilege escalation
- Checking sudo permissions
Checking sudo permissions is listed as a task the coordinator must delegate rather than perform directly. In context, it supports privilege-escalation enumeration against a compromised target.
High
SKILL.md:41
sudo privilege escalation
prompt="Escalate privileges from www-data to root. Check sudo -l, find SUID binaries, check capabili
The prompt directly instructs escalation from www-data to root using sudo -l, SUID binaries, capabilities, and linpeas. This is explicit privilege-escalation guidance.
High
SKILL.md:25
Hardcoded IP address
prompt="Perform comprehensive reconnaissance on target 10.10.10.1. Scan all ports, enumerate service
The prompt tells a subagent to scan all ports, enumerate services, and check web directories on target 10.10.10.1. This is direct network reconnaissance guidance.
High
SKILL.md:125
Hardcoded IP address
3. Delegate: Task(subagent_type="Explore", prompt="Scan 10.10.10.1...")
The coordinator delegates a scan against 10.10.10.1 in the example flow. That is an actionable network-scanning instruction, not only a harmless literal IP address.

Risk Factors

⚙️ External commands (13)
SKILL.md:21-45 SKILL.md:45-50 SKILL.md:50-52 SKILL.md:52-56 SKILL.md:56-76 SKILL.md:76-95 SKILL.md:95-98 SKILL.md:98-119 SKILL.md:119-156 SKILL.md:156-170 SKILL.md:170-177 SKILL.md:17 SKILL.md:41
🌐 Network access (3)
SKILL.md:25 SKILL.md:120 SKILL.md:125

Detected Patterns

Network reconnaissanceNetwork scanning tools
Audited by: codex View Audit History →
Share & cite this report

Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.

Open versioned report
Security Assessment

Copy report link

https://skillstore.io/skills/charleskozel-coordinator/audits/8?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown badge

[![Skillstore security assessment](https://skillstore.io/badges/skills/charleskozel-coordinator/security.svg)](https://skillstore.io/skills/charleskozel-coordinator?utm_source=security_passport_badge)

HTML badge

<a href="https://skillstore.io/skills/charleskozel-coordinator?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/charleskozel-coordinator/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Embed card

<iframe src="https://skillstore.io/embed/skills/charleskozel-coordinator.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>
Academic citations (APA · BibTeX · CFF)

APA citation

CharlesKozel. (2026). coordinator security audit report (audit version 8) [Author version unspecified]. Skillstore. https://skillstore.io/skills/charleskozel-coordinator/audits/8

BibTeX citation

@techreport{charleskozel-charleskozel-coordinator-2026, author = {CharlesKozel}, title = {coordinator security audit report (audit version 8)}, institution = {Skillstore}, year = {2026}, number = {8}, url = {https://skillstore.io/skills/charleskozel-coordinator/audits/8}, note = {Author version unspecified} }

CITATION.cff

cff-version: 1.2.0 message: "If you use this Skill, cite its author and this versioned security audit report." title: "coordinator security audit report (audit version 8)" version: "unspecified" type: report authors: - name: "CharlesKozel" date-released: "2026-07-06" url: "https://skillstore.io/skills/charleskozel-coordinator/audits/8" identifiers: - type: other value: "skillstore:charleskozel-coordinator:audit:8" description: "Skillstore immutable audit report identifier"

Skillstore Score

Why this score Evidence Confidence: Medium
55
Architecture
85
Maintainability
87
Content
70
Community
83
Spec Compliance

What You Can Build

Coordinate CTF Lab Progress

Track phases and delegate next steps during an isolated capture-the-flag exercise.

Run Internal Range Exercises

Structure agent handoffs for a sanctioned training range with predefined targets and rules.

Review Agentic Security Workflows

Study how autonomous security agents coordinate tasks, state, and completion criteria.

Try These Prompts

Start Authorized Lab 
Coordinate an authorized CTF assessment for TARGET_LAB_IP. Read the current phase, assign reconnaissance, and summarize safe next steps.
Resume From State 
Review the current lab state, identify completed phases, and recommend the next delegated task within the approved scope.
Assess Agent Findings 
Analyze the latest agent findings from an authorized lab. Decide whether to continue enumeration, test a finding, or document results.
Constrain AD Lab Work 
Coordinate an Active Directory training lab with strict scope, stopping rules, and evidence-based task delegation only.

Best Practices

  • Use only in isolated labs, CTFs, or systems with written authorization.
  • Require explicit target scope, allowed techniques, and stopping rules before activation.
  • Review each delegated action before execution when real systems may be reachable.

Avoid

  • Do not run it against public IP addresses or third-party networks.
  • Do not keep non-termination rules that prevent stopping or user review.
  • Do not treat generated exploit plans as proof of authorization or accuracy.

Frequently Asked Questions

Is this skill safe for production networks?
No. It should be limited to authorized labs or isolated training ranges.
Does it execute scans itself?
The coordinator delegates scanning and exploitation tasks to other agents. Those delegated actions can still create real network risk.
What tools does it support?
The report lists Claude, Codex, and Claude Code compatibility.
Can it verify authorization?
No. Users must provide scope and legal approval before any testing.
Why is the audit high risk?
The source directs autonomous reconnaissance, exploitation, privilege escalation, and non-stop behavior until flags are captured.
How should authors make it safer?
Remove autonomous exploitation, add consent checks, require stop conditions, and keep guidance educational unless lab scope is explicit.

Developer Details

License

MIT

Skillstore revision

r1

Version notice

The author did not declare a version.

Repository

https://github.com/CharlesKozel/Pentest-Agent-Evalulator/tree/main/agents/claude-tbug/skills/coordinator

Ref

dd4a3ef9f20ddf38830950b4bb713df96b431fd6

Maintenance freshness

7/22/2026

Usage

7 downloads · 203 views

File structure

📄 AD_ATTACK_SUPPLEMENT.md

📄 SKILL.md