agentic-threat-hunting-framework 0.2.3__py3-none-any.whl → 0.2.4__py3-none-any.whl

athf/data/prompts/README.md

@@ -0,0 +1,172 @@
+# AI Prompt Library
+
+This folder contains prompts to help you accelerate threat hunting at different maturity levels.
+
+---
+
+## What's Here
+
+### basic-prompts.md
+
+**Level:** 0-1 (Manual/Documented)
+**Use for:** Copy-paste prompts for ChatGPT, Claude, or other AI assistants
+
+Contains three prompt templates:
+
+1. **Generate Hypothesis** - From CTI, alerts, or anomalies
+2. **Build Query** - Safe, bounded queries for Splunk, KQL, or Elastic
+3. **Document Results** - Capture findings in LOCK format
+
+**When to use:** You're working outside an AI-enabled IDE and need quick assistance with hypothesis generation, query building, or documentation.
+
+---
+
+### ai-workflow.md
+
+**Level:** 2 (Searchable) - AI with Memory
+**Use for:** AI tools that can read your repository (Claude Code, GitHub Copilot, Cursor)
+
+Contains:
+
+- System prompt for AI tools
+- 4 core workflows (threat intel, anomaly investigation, proactive hunting, documentation)
+- Complete example conversation showing AI reasoning
+- Tool-specific tips and troubleshooting
+- Quality checklists
+
+**When to use:** You have AI tools with file access to your hunt repository and want them to search past hunts, validate against environment.md, and generate context-aware hypotheses.
+
+---
+
+## How to Choose
+
+**Use basic-prompts.md if:**
+
+- You're just getting started with AI-assisted hunting
+- You don't have AI tool subscriptions yet
+- You want simple copy-paste templates
+- You're working in a web interface (ChatGPT, Claude.ai)
+
+**Use ai-workflow.md if:**
+
+- You have Claude Code, GitHub Copilot, or Cursor
+- Your hunt repository has AGENTS.md, knowledge/hunting-knowledge.md, and documented past hunts
+- You want AI to search memory, apply expert hunting frameworks, and apply lessons learned
+- You're ready for more advanced workflows
+
+---
+
+## Quick Start
+
+### Level 0-1: Basic Prompts
+
+1. Open [basic-prompts.md](basic-prompts.md)
+2. Copy the prompt template you need
+3. Fill in your context (hypothesis, data sources, results)
+4. Paste into ChatGPT, Claude, or your AI assistant
+5. Review and refine the output
+
+**Example:**
+
+```
+# You have threat intel about PowerShell abuse
+→ Use "Generate Hypothesis" prompt from basic-prompts.md
+→ Paste CTI report into context section
+→ AI generates testable hypotheses
+```
+
+### Level 2: AI Workflows
+
+1. Open your hunt repository in Claude Code, Copilot, or Cursor
+2. Provide the system prompt from [ai-workflow.md](ai-workflow.md)
+3. Ask AI to search past hunts before generating new ones
+4. Follow the workflow guides for common scenarios
+
+**Example:**
+
+```
+You: "Check if we've hunted T1003.001 before. Use the system prompt from prompts/ai-workflow.md"
+AI: [Searches hunts/, reads environment.md, generates context-aware hypothesis]
+```
+
+---
+
+## Safety Reminders
+
+### AI Assistance ≠ Autopilot
+
+- **Always review** AI-generated hypotheses for feasibility
+- **Always test** AI-generated queries on small timeframes first
+- **Always validate** that queries are safe and bounded
+- **Use your judgment** - You know your environment better than AI
+
+### Before Running Any AI-Generated Query
+
+1. Check for time bounds (`earliest=-Xd`)
+2. Check for result limits (`| head N` or `| take N`)
+3. Test on 1-hour window before expanding to days
+4. Verify it won't impact SIEM performance
+
+---
+
+## Platform-Specific Tips
+
+**Splunk Users:**
+
+- Mention "Splunk SPL" in your prompts
+- Specify data models when available
+- AI knows common Splunk patterns (tstats, eval, stats)
+
+**KQL Users (Sentinel/Defender):**
+
+- Mention "KQL for Sentinel" or "KQL for Defender"
+- Specify table names (SecurityEvent, DeviceProcessEvents, etc.)
+- AI understands Sentinel-specific syntax
+
+**Elastic Users:**
+
+- Mention "Elastic EQL" or "Lucene query"
+- Specify index patterns
+- Note which Elastic stack version you're using
+
+---
+
+## Next Steps
+
+### After Using Basic Prompts
+
+1. Document your hunts using [templates/HUNT_LOCK.md](../templates/HUNT_LOCK.md)
+2. Create AGENTS.md in your repository (see main README)
+3. Ensure knowledge/hunting-knowledge.md is present (included in repo by default)
+4. Progress to Level 2 with ai-workflow.md
+
+### After Level 2 Workflows
+
+1. See real examples in [hunts/H-0001.md](../hunts/H-0001.md) and [hunts/H-0002.md](../hunts/H-0002.md)
+2. Review format guidelines in [hunts/FORMAT_GUIDELINES.md](../hunts/FORMAT_GUIDELINES.md)
+3. Consider Level 3 (MCP integrations) in [integrations/](../integrations/)
+
+---
+
+## Customizing for Your Environment
+
+Feel free to modify these prompts:
+
+- Add your organization's specific data sources
+- Include your ATT&CK coverage gaps
+- Reference your baseline automation
+- Add your threat model priorities
+
+---
+
+## Contributing
+
+Have a better prompt? Found a useful workflow?
+
+- Submit a PR with your improvements
+- Share what works in your environment
+- Help others get started faster
+
+---
+
+**Remember: These prompts are training wheels. They help you get started faster, teach you the LOCK pattern, and over time you'll need them less. But they remain useful for complex hunts.**