Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/briiirussell/cybersecurity-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/rules/briiirussell/cybersecurity-skills/threat-hunting)<a href="https://agentmods.dev/rules/briiirussell/cybersecurity-skills/threat-hunting"><img src="https://agentmods.dev/badge/rules/briiirussell/cybersecurity-skills/threat-hunting/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/rules/briiirussell/cybersecurity-skills/threat-hunting"><img src="https://agentmods.dev/badge/rules/briiirussell/cybersecurity-skills/threat-hunting.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00123 | $0.02906 |
| Opus 5.5 | $0.00049 | $0.01162 |
| Sonnet 5.5 | $0.00025 | $0.00581 |
| Haiku 4.5 | $0.00012 | $0.00291 |
Grade C, and why
threat-hunting scanned grade C with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured today.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Cloud metadata endpointhighServer-side request forgery
One request to 169.254.169.254 can return temporary IAM credentials.
- **IMDS access from unusual processes / containers** — anything reaching `169.254.169.254` that isn't the cloud SDK How it starts
The opening of the file, as written. The whole thing — 230 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Threat Hunting — Proactive Adversary Detection
Hunt for adversaries who are already inside but haven't tripped an alert. Distinct from incident-triage (reactive, alert is firing) and from siem-detection (engineer rules so future alerts fire). This skill is the proactive layer — assume something has slipped through, look for it.
Hunting is hypothesis-driven, not browse-driven. "Let's look around the SIEM" is not hunting; "let's check for the specific pattern of T1059.001 (PowerShell) being launched by Office processes" is.
Cross-references: siem-detection (queries you write here often graduate to detection rules), incident-triage (what to do if a hunt confirms a finding), breach-patterns (a rich source of hunt hypotheses), disk-forensics (deeper analysis on confirmed hits), agent-supply-chain-audit (hunting for malicious or rug-pulled AI coding-agent extensions on developer machines), secrets-audit (its exfiltration-collector domain list is a ready-made egress hunt: DNS / proxy hits to webhook.site, requestbin, ngrok, or chat webhooks from servers or CI runners).
Authorization
Hunt only in environments the user owns or is authorized to investigate, with the logs and telemetry they've given you access to. Hunting reads; it doesn't contain. If a hunt finds confirmed malicious activity, stop and hand off to incident-triage instead of acting on hosts or accounts. Hunting in employee activity stays within what HR and legal have authorized.
Methodology — PEAK framework
The PEAK (Prepare, Execute, Act, Knowledge) framework from Splunk SURGe — the most actionable hunting methodology I've seen.
Step 1: Prepare
Form the hypothesis. Strong hypotheses share three properties:
- Specific — names a technique, log source, and expected artifact
- Testable — describes what evidence would confirm or deny
- Bounded — has a defined time window and scope
Bad hypothesis: "Look for anomalies in the SIEM"
Good hypothesis: "Within the last 30 days, no service account should have run interactive PowerShell with -encodedCommand flag (T1059.001 + T1027). Search Sysmon event 1 for parent process = service-account-launched scheduled task, child = powershell.exe, command line contains -enc or -encodedcommand."
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- today Changed · +4 lines 18a63d8b5da0
- 1mo ago First seen · 226 lines · 123 tokens per session scan C 1f975dea9832
threat-hunting is a cursor rule published in the GitHub repository briiirussell/cybersecurity-skills (435 stars, last pushed today), licensed MIT. It adds 123 tokens to every session and 2,906 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it C with 1 finding (cloud metadata endpoint). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other cursor rules, from other repositories
ponytail
Ponytail, lazy senior dev mode. Always pick the simplest solution that works.
angular-20
This rule provides comprehensive best practices and coding standards for Angular development, focusing on modern TypeScript, standalone components, signals, and performance optimizations.
dev-standard
Apache Superset development standards and guidelines for Cursor IDE.
control-plane-descriptors
Control plane descriptor and instance implementation patterns.
family-instance-domain-actions
Family instance domain action implementation patterns.
no-family-vocabulary-in-framework
The framework domain (packages/1-framework) carries no family- or target-specific vocabulary — types, fields, hooks, or strategy values — and, outside the CLI and the PSL parser's grammar names, no Prisma version name before 8. Enforced by the no-family-vocabulary Biome plugin plus the lint:ratchets ratchet.