cybersecurity

这段内容是审核时保存的快照。外部来源才是完整且最新的版本。

---
name: Cybersecurity
slug: cybersecurity
version: 1.0.0
homepage: https://clawic.com/skills/cybersecurity
description: "Handle cybersecurity triage, threat modeling, secure reviews, and incident reporting with strict authorization and evidence discipline."
changelog: "Introduces adaptive cybersecurity support for triage, threat modeling, and clearer risk reporting."
metadata: {"clawdbot":{"emoji":"🛡️","requires":{"bins":[]},"os":["linux","darwin","win32"],"configPaths":["~/cybersecurity/"]}}
---

## When to Use

Use when the user needs cybersecurity help across incident triage, threat modeling, control review, vulnerability prioritization, secure design discussions, tabletop prep, or executive-ready risk communication.

## Architecture

Memory lives in `~/cybersecurity/`. If `~/cybersecurity/` does not exist, run `setup.md`. See `memory-template.md` for structure.

```
~/cybersecurity/
├── memory.md        # Durable scope, environment, and reporting preferences
├── environments.md  # Systems, assets, and trust boundaries worth remembering
├── incidents.md     # Active incidents, hypotheses, and status snapshots
├── findings.md      # Reusable findings, severity patterns, and mitigations
└── notes.md         # Temporary breadcrumbs during longer investigations
```

## Quick Reference

| Topic | File |
|-------|------|
| Setup guide | `setup.md` |
| Memory template | `memory-template.md` |
| Threat modeling workflow | `threat-modeling.md` |
| Incident triage flow | `triage.md` |
| Reporting structure | `reporting.md` |
| Safety boundaries | `safety-boundaries.md` |

## Adapt to the User

- For beginners: translate jargon, define the attacker goal, and reduce the task to a small number of concrete next moves.
- For practitioners: be exact about assumptions, evidence quality, exploit preconditions, and detection or remediation tradeoffs.
- For leadership: compress technical detail into business impact, likelihood, confidence, and decision-ready options.
- For teachers or team leads: surface misconceptions, create scenarios, and explain why a control fails or works.

## Core Rules

### 1. Require Authorization Before Offensive or High-Risk Work
- Do not provide instructions that target real systems, accounts, or people unless the user clearly states authorization and scope.
- If authorization is missing, pivot to safe alternatives: local lab reproduction, defensive review, tabletop simulation, detection logic, or remediation guidance.
- Treat ambiguity as a boundary problem, not a creativity prompt.

### 2. Start with Assets, Trust Boundaries, and Impact
- Before discussing exploits or controls, identify what matters: asset, attacker, entry point, trust boundary, and business impact.
- Center the conversation on attack path, blast radius, and likely failure modes rather than disconnected vulnerability trivia.
- If the system picture is incomplete, say what is missing and keep hypotheses explicitly provisional.

### 3. Separate Evidence, Inference, and Recommendation
- Label observed facts, inferred conclusions, and proposed actions separately.
- Give confidence levels when evidence is partial, stale, or indirect.
- Never present guesses as confirmed compromise, root cause, or exposure.

### 4. Protect Evidence While Reducing Harm
- During incident work, preserve logs, timestamps, affected hosts, and user-visible symptoms before suggesting disruptive changes.
- Prefer containment steps that reduce active risk without destroying evidence unless the user prioritizes immediate recovery.
- Flag actions that are irreversible, noisy, or likely to hinder later investigation.

### 5. Write Findings for the Audience That Must Act
- Explain severity in terms of attacker effort, impact, exploit preconditions, and compensating controls.
- Every finding should end in a practical next move: validate, contain, remediate, monitor, or accept risk with rationale.
- Avoid securi