Repository navigation
AnythingLLM Agentflow Documentation
Custom AgentFlow configurations for cybersecurity assessments, penetration testing, and security automation
Part of the ULTIMATE CYBERSECURITY MASTER GUIDE
Agent acceptance tests: Promptfoo security evaluation workflow covers synthetic data, tool permissions and instruction boundaries before connecting customer data.
Index and full technical documentation for 9 custom AnythingLLM AgentFlows covering OSINT/recon, scanning analysis, and vulnerability reporting. Complements anything_llm_agentflows.md, which shows how to chain these flows together into multi-step workflows rather than documenting each flow's internal block configuration.
A categorized flow catalog (OSINT/Recon, Scanning/Analysis, Vulnerability Management) with quick-reference tables by use case, complexity, and required API keys, followed by each flow's dedicated flow-*.md file containing exact block-by-block configuration for recreating it in AnythingLLM.
Recreate any of the 9 documented AgentFlows in your own AnythingLLM instance from the block-level configuration, and know which flow to use for a given engagement phase (pre-engagement, active testing, reporting).
- Setting up AnythingLLM AgentFlows for the first time (start here, then open the specific
flow-*.md) - Deciding which flow fits a given engagement phase (see "By Use Case" table)
- Troubleshooting a flow that isn't producing expected output (see "Common Issues & Solutions")
This directory contains complete technical documentation for creating custom AgentFlows used in cybersecurity assessments and penetration testing. Each flow is documented with full configuration details, block specifications, and usage instructions.
Purpose: Gathers public information about target companies
Outputs: Email formats, employees, tech stack, social media presence
Use Cases: Pre-engagement intelligence, company profiling
Blocks: 6 (3 web scraping, 3 LLM)
Purpose: Performs comprehensive domain reconnaissance
Outputs: Subdomains, DNS records, certificates, security posture
Use Cases: Attack surface mapping, asset discovery
Blocks: 6 (4 web scraping, 1 LLM, 1 complete)
Purpose: Investigates email addresses and usernames across platforms
Outputs: Breach history, social media accounts, security awareness assessment
Use Cases: Social engineering prep, breach verification
Blocks: 6 (1 API call, 1 web scraping, 3 LLM, 1 complete)
Requirements: HaveIBeenPwned API key
Purpose: Searches GitHub for exposed credentials and sensitive data
Outputs: Exposed secrets, API keys, risk assessment
Use Cases: Credential exposure detection, pre-engagement OSINT
Blocks: 5 (2 web scraping, 2 LLM, 1 complete)
Purpose: Checks domains/emails against known data breaches
Outputs: Breach history, exposed data types, risk analysis
Use Cases: Domain-wide breach assessment, credential exposure
Blocks: 4 (2 API calls, 1 LLM, 1 complete)
Requirements: HaveIBeenPwned API key
Purpose: Analyzes Nmap scan results and provides recommendations
Outputs: Parsed findings table, prioritized risks, next steps
Use Cases: Port scan analysis, vulnerability prioritization
Blocks: 3 (2 LLM, 1 complete)
Purpose: Checks IP/domain reputation against threat intelligence
Outputs: Threat score, BLOCK/MONITOR/ALLOW recommendation
Use Cases: IP reputation checking, incident response
Blocks: 5 (1 API call, 1 web scraping, 2 LLM, 1 complete)
Requirements: VirusTotal API key (recommended)
Purpose: Retrieves CVE details and exploitability information
Outputs: CVSS analysis, exploitability assessment, remediation plan
Use Cases: CVE research, vulnerability analysis
Blocks: 4 (1 web scraping, 2 LLM, 1 complete)
Purpose: Creates professional vulnerability reports
Outputs: Executive summary, technical findings, remediation steps
Use Cases: Report writing, client deliverables
Blocks: 5 (4 LLM, 1 complete)
Pre-Engagement Phase:
- CompanyOSINT
- DomainRecon
- EmailOSINT
- GitHubSecrets
- BreachChecker
Active Testing Phase:
- NmapAnalyzer
- ThreatIntelCheck
- CVELookup
Reporting Phase:
- VulnReportGenerator
Simple (3-4 blocks):
- NmapAnalyzer
- CVELookup
- BreachChecker
Medium (5-6 blocks):
- GitHubSecrets
- ThreatIntelCheck
- EmailOSINT
- DomainRecon
- VulnReportGenerator
Complex (6+ blocks):
- CompanyOSINT
Essential:
-
HaveIBeenPwned API Key - Required for EmailOSINT and BreachChecker
- Get at: https://haveibeenpwned.com/API/Key
- Cost: $3.50/month (as of 2025)
Recommended:
-
VirusTotal API Key - Required for ThreatIntelCheck
- Get at: https://www.virustotal.com/gui/join-us
- Free tier: 4 req/min, 500/day
Optional:
-
AbuseIPDB API Key - Improves ThreatIntelCheck
- Get at: https://www.abuseipdb.com/register
- Free tier: 1,000 checks/day
-
WhoisXML API Key - Improves DomainRecon
-
SecurityTrails API Key - Improves DomainRecon
/mnt/user-data/outputs/
โโโ anythingllm-workflow-examples.md # Workflow combinations guide
โโโ flow-CompanyOSINT.md # Company intelligence
โโโ flow-DomainRecon.md # Domain reconnaissance
โโโ flow-EmailOSINT.md # Email investigation
โโโ flow-GitHubSecrets.md # GitHub credential scanning
โโโ flow-BreachChecker.md # Breach verification
โโโ flow-NmapAnalyzer.md # Nmap scan analysis
โโโ flow-ThreatIntelCheck.md # IP/domain reputation
โโโ flow-CVELookup.md # CVE intelligence
โโโ flow-VulnReportGenerator.md # Report generation
โโโ README-AgentFlows.md # This file
- Choose a flow from the list above
- Open the corresponding .md file
- Follow the "Flow Variables" section - Set up all variables
- Follow the "Flow Blocks" section - Create each block in order
- Configure each block exactly as documented
- Save and publish the flow
- Test using examples in "Usage Instructions"
- Quick lookup of flow configuration
- Troubleshooting common issues
- API setup instructions
- Integration patterns with other flows
- Each flow can be customized
- Add additional data sources
- Modify prompts for different output formats
- Chain with custom flows
Problem: Variables show as {{variable}} instead of values
Solution: Use ${variable} syntax in URLs and prompts
Problem: 401/403 errors from APIs
Solution: Verify API keys are correct and active
Problem: "Quota exceeded" or "Too many requests"
Solution: Wait between requests, upgrade to paid tiers
Problem: Scraping returns errors or empty data
Solution: Check if site blocks bots, consider API alternatives
Problem: "Result too long, truncated"
Solution: Enable content summarization, break into smaller queries
- Start Simple - Create basic flow, test, then enhance
- Test Each Block - Verify each block works before adding next
- Use Meaningful Names - Clear variable and block names
- Document As You Go - Note any customizations made
- Version Control - Keep notes on flow versions and changes
- Verify Authorization - Always have permission before scanning
- Respect Rate Limits - Don't exceed API quotas
- Validate Results - Cross-check critical findings
- Document Findings - Keep records of all flows run
- Secure API Keys - Never commit keys to repositories
- Chain Flows Logically - Follow reconnaissance โ analysis โ reporting
- Save Intermediate Outputs - Useful for troubleshooting
- Batch Processing - Run similar flows together
- Automate Common Tasks - Script repetitive flow usage
- Quarterly Review - Check if APIs still work
- Update API Keys - Rotate keys as needed
- Refresh Prompts - Improve based on output quality
- Add New Sources - Incorporate new threat intel feeds
- Track Success Rate - Note failed flows
- Monitor API Quotas - Avoid exhausting limits
- Review Output Quality - Ensure prompts still effective
- User Feedback - Improve based on actual usage
- Each flow .md file contains complete technical details
- Workflow examples:
anythingllm-workflow-examples.md - Usage patterns and best practices included
- AnythingLLM Docs: Check official documentation for updates
- API Documentation: Refer to provider docs for API changes
- Security Communities: OSINT forums, security subreddits
- Check individual flow documentation "Troubleshooting" section
- Review "Known Issues" sections for common problems
- Test with example inputs provided in each flow doc
Documentation Version: 1.0
Last Updated: November 2025
Created By: Pacific Northwest Computers
Purpose: Complete AgentFlow recreation guide
When creating new flows, document:
- Flow purpose and use case
- All variables with descriptions
- Each block configuration with exact settings
- Complete prompts and instructions
- Usage examples
- Integration patterns
- Troubleshooting tips
When modifying flows:
- Update corresponding .md file
- Note version changes
- Document what changed and why
- Update usage examples if needed
- Test thoroughly before documenting
Purpose: Internal use for cybersecurity assessments
Authorization: Only use flows on authorized targets
Compliance: Follow all applicable laws and regulations
Ethical Use: Responsible disclosure of all findings
- ../README.md - AI section index
- anything_llm_agentflows.md - Chained workflow examples using these flows
- flow-breacher_check.md - Breach/credential leak checker flow
- flow-company_osint.md - Company OSINT automation flow
- flow-cve_lookup.md - CVE research and analysis flow
- flow-domain_recon.md - Domain reconnaissance flow
- flow-email_osint.md - Email OSINT investigation flow
- flow-nmap_analyzer.md - Nmap output analysis flow
- flow-threat_intel.md - Threat intelligence lookup flow
- flow-vulnreport_gen.md - Vulnerability report generation flow
- ../offline-llm.md - Offline LLM backend that AnythingLLM can use
๐ Source: AI/AnythingLLM/README.md | โ๏ธ Suggest an edit | โฌ
๏ธ Back to Artificial Intelligence
- ๐บ๏ธ Browse by Topic
- ๐ All Repository Files
- ๐งญ Learning Paths
- ๐ Master Guide
- ๐ค Glossary
- โ๏ธ Legal & Disclaimer
๐ Foundations & Learning
๐ก๏ธ Defense & Investigations
- Incident Response
- Open Source Intelligence (OSINT)
- Reverse Engineering
- Operations Security (OPSEC)
- Compliance
โ๏ธ Platforms & Infrastructure
๐ก Hardware, Wireless & Field Projects
- PNWC Cyberdeck
- Hardware Testing
- Hardware Hacking
- Software-Defined Radio
- Wardriving
- uConsole
- Space Security
๐งฐ Assessment & Field Reference
๐ฆ Downloads & Project Resources
๐๏ธ Additional Topics