Skills DirectorySearch all skills ↗AI agent skill · by @brucesongs
Article Writing
Transform technical findings into clear, structured written content: penetration test reports, vulnerability disclosures, security blog posts, and technical documentation.
Archived 2026-10-23 · Source updated 2026-09-09 · License: MIT. Check the original repository for the latest version.
SKILL.md instructions
# Article Writing ## Summary Article Writing skill domain covering knowledge operations. **Domain**: knowledge ## Skill Identity | Attribute | Value | |-----------|-------| | Domain | Documentation & Communication | | Skill ID | article-writing | | Version | 1.0.0 | | Hacker Laws | Law 8 (Learn from Every Operation), Law 11 (Share Knowledge, Build Legacy) | | Related Skills | knowledge-ops, deep-research | ## Purpose Transform technical findings into clear, structured written content: penetration test reports, vulnerability disclosures, security blog posts, and technical documentation. Writing is the final step in the intelligence workflow — turning raw findings into actionable deliverables for clients, researchers, or the public. ## Article Types | Type | Audience | Format | Length | |------|----------|--------|--------| | **Pentest Report** | Client (technical + executive) | Executive Summary + Technical Findings | 10-50 pages | | **Vulnerability Disclosure** | Vendor security team | CVE template + PoC | 2-5 pages | | **Security Blog Post** | Public (technical) | Narrative + code samples | 1000-3000 words | | **Advisory** | Public (mixed audience) | CVSS + mitigation steps | 500-1500 words | | **Technical Documentation** | Internal team | How-to guide | Variable | ## Methodology ### Phase 1: Gather Intelligence ```bash # Aggregate all findings from knowledge-ops grep -rn "type: finding" memory/*target*.md | wc -l # Extract high-confidence findings (>= 75) grep -rn "confidence: [789][0-9]\|confidence: 100" memory/*target*.md -l | \ xargs grep -h "## Summary" # Group by severity grep -rn "tags:.*critical\|tags:.*high" memory/*target*.md ``` ### Phase 2: Structure Content Follow the appropriate template (see below) for the article type. Outline before writing. ### Phase 3: Write - **Executive Summary**: Non-technical, business impact-focused - **Technical Detail**: Reproducible, includes code/commands - **Evidence**: Screenshots, logs, PoC code - **Recommendations**: Actionable remediation steps ### Phase 4: Review - Technical accuracy check - No sensitive data leakage (sanitize IPs, domains, credentials) - CVSS scoring (if applicable) - Proofread for clarity ## Templates ### Pentest Report Template ```markdown # Penetration Test Report: [Client Name] **Date**: [date] **Version**: [version] **Prepared by**: [your org] **Classification**: [Confidential/Restricted] --- ## Executive Summary [Target]: [description of target scope] [Duration]: [test dates] [Findings Summary]: [count by severity] ### Key Findings - **Critical**: [count] — [one-line impact] - **High**: [count] — [one-line impact] - **Medium**: [count] — [one-line impact] - **Low**: [count] — [one-line impact] ### Business Impact [2-3 sentences on overall risk to the organization] ### Recommendations Priority 1. [Most critical fix] 2. [Second priority] 3. [Third priority] --- ## Scope **In Scope**: - [systems/domains tested] **Out of Scope**: - [systems excluded] **Testing Methodology**: [OWASP, PTES, custom] --- ## Findings ### Finding 1: [Vulnerability Name] **Severity**: Critical **CVSS**: 9.8 (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H) **Affected Systems**: [list] ## ## Description [What is the vulnerability? How does it work?] #### Evidence ```bash [Commands used to discover/exploit] ``` [Screenshot or log output] #### Impact - **Confidentiality**: [High/Medium/Low] — [why] - **Integrity**: [High/Medium/Low] — [why] - **Availability**: [High/Medium/Low] — [why] #### Remediation **Short-term**: [immediate fix] **Long-term**: [architectural improvement] **Verification**: [how to verify fix] --- [Continue for all findings] --- ## Appendix ### Tools Used - [tool list with versions] ### Testing Timeline | Date | Activity | |------|----------| | [date] | Reconnaissance | | [date] | Vulnerability scanning | | [date] | Manual testing | | [date] | Exploitation | | [date] | Reporting | ### References - [OWASP guides, CVE references, etc.] ``` ### Vulnerability Disclosure Template ```markdown # Vulnerability Disclosure: [Vulnerability Name] **Date**: [discovery date] **Severity**: [Critical/High/Medium/Low] **CVSS**: [score] ([vector]) **CVE ID**: [if assigned, otherwise "Pending"] --- ## Affected Products - **Product**: [name] - **Versions**: [affected versions] - **Fixed in**: [patched version, if known] ## Vulnerability Details ### Type [SQL Injection / XSS / SSRF / etc.] ### Location - **File**: [file path or URL] - **Parameter**: [vulnerable parameter] - **Method**: [GET/POST/etc.] ### Root Cause [Technical explanation of why the vulnerability exists] ## Proof of Concept ```bash [Minimal PoC to reproduce] ``` **Expected Result**: [what should happen] **Actual Result**: [what happens, demonstrating the vuln] ## Impact [Detailed impact analysis — what can an attacker achieve?] ## Remediation ### Developer Fix ```diff [Code diff showing the fix, if simple] ``` ### Workaround (if patch not available) [Temporary mitigation steps] ## Timeline | Date | Event | |------|-------| | [date] | Vulnerability discovered | | [date] | Vendor notified | | [date] | Vendor acknowledged | | [date] | Patch released | | [date] | Public disclosure | ## Credits [Your name/org] ## Detection Methods ### Content Authenticity Audit - **AI-generated text detection**: Statistical analysis (perplexity, burstiness) for AI fingerprints. - **Plagiarism detection**: Content matching against existing web corpus (Copyscape, Turnitin). - **Style anomalies**: Sudden style change in author's corpus; co-author compromise. ### SIEM Detection Rules - **Custom content analytics**: Per-author baseline; alert on deviation. ## Defense Evasion Techniques ### AI Content Stealth - **Humanize AI output**: Use Wordtune, Quillbot to add human-like variance. - **Mixed authorship**: Combine AI-generated sections with human-written sections; reduces perplexity anomaly. - **Style mimicry**: Train model on target author's corpus; mimic their style. ## References - [Related CVEs, vendor advisories, etc.] ``` ### Security Blog Post Template ```markdown # [Catchy Title]: [One-Sentence Hook] **Published**: [date] **Author**: [name] **Tags**: [tag1, tag2, tag3] --- ## TL;DR [2-3 sentences summarizing the entire post] --- ## Introduction [Set the stage — why does this topic matter? What problem are you solving?] ## Background [Context needed to understand the vulnerability/technique/tool] ## Discovery [How did you find this? What were the initial signals?] ## Deep Dive [Technical details — this is the meat of the post] ### Step 1: [Phase Name] [Description] ```bash [Commands] ``` [Output/screenshot] ### Step 2: [Phase Name] [Continue for all steps] ## Impact & Exploitation [What can an attacker do with this? Real-world scenarios] ## Detection & Mitigation ### For Defenders - [Detection method 1] - [Detection method 2] ### For Developers - [Secure coding practice to prevent this] ## Conclusion [Wrap up — key takeaways, call to action] ## References - [Links to related research, tools, CVEs] ``` ## Use Cases 1. **Pentest Deliverable**: Final report for client after engagement 2. **Responsible Disclosure**: Notify vendor of discovered vulnerability 3. **Knowledge Sharing**: Publish research findings publicly 4. **Internal Documentation**: Record methodology for team playbooks 5. **Advisory Publishing**: Warn community of active threat ## Writing Best Practices - **Be precise**: Use exact file paths, line numbers, version numbers - **Be reproducible**: Include all commands, payloads, and environment details - **Be cautious**: Sanitize sensitive data (IPs, domains, real credentials) - **Be visual**: Screenshots, diagrams, code blocks - **Be actionable**: Every finding needs a clear fix ## Integration - **Input**: knowledge-ops findings, test-cases validation results - **Output**: Markdown/PDF reports, blog posts, advisories - **Handoff**: Share with client, vendor, or publish publicly