docs: add README with usage guide and integration examples

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
salvacybersec
2026-03-22 01:11:49 +03:00
parent 349fcd6d4b
commit 9c9d1fb8d6

254
README.md Normal file
View File

@@ -0,0 +1,254 @@
# Persona Library
Platform-agnostic LLM persona library for AI agents. 29 personas across 10 domains with 25+ specialization variants.
## Quick Start
```bash
# Generate all formats (.yaml, .json, .prompt.md)
python3 build.py
# Use a persona as system prompt
cat .generated/neo/general.prompt.md
# Use structured data for API integration
cat .generated/frodo/iran.json
```
## Domains & Personas
### Cybersecurity (7)
| Codename | Role | Hitap | Variants |
|----------|------|-------|----------|
| **Neo** | Red Team Lead / Exploit Dev | Sıfırıncı Gün | general, redteam, exploit-dev, wireless |
| **Phantom** | Web App Security / Bug Bounty | Beyaz Şapka | general, api-security |
| **Cipher** | Cryptography & Crypto Analysis | Kriptoğraf | general |
| **Specter** | Malware Analyst / Reverse Engineer | Cerrah | general |
| **Bastion** | Blue Team / DFIR | Muhafız | general, forensics, threat-hunting |
| **Vortex** | Network Ops / Traffic Analysis | Telsizci | general, cloud-ad |
| **Sentinel** | CTI / Threat Intelligence | İzci | general, apt-profiling, mitre-attack |
### Intelligence (5)
| Codename | Role | Hitap | Variants |
|----------|------|-------|----------|
| **Frodo** | Strategic Intelligence Analyst | Müsteşar | general, middle-east, russia, iran, africa, china |
| **Oracle** | OSINT & Digital Intelligence | Kaşif | general |
| **Ghost** | PSYOP & Information Warfare | Propagandist | general, cognitive-warfare |
| **Wraith** | HUMINT & Counter-Intelligence | Mahrem | general, source-validation |
| **Echo** | SIGINT / COMINT / ELINT | Kulakçı | general, nsa-sigint |
### Military & Strategy (4)
| Codename | Role | Hitap | Variants |
|----------|------|-------|----------|
| **Marshal** | Military Doctrine & Strategy | Mareşal | general, nato-doctrine, hybrid-warfare |
| **Warden** | Defense Analyst / Weapons Systems | Topçubaşı | general |
| **Centurion** | Military History & War Analysis | Vakanüvis | general |
| **Corsair** | Special Operations & Irregular Warfare | Akıncı | general |
### Law, Economics & Politics (3)
| Codename | Role | Hitap | Variants |
|----------|------|-------|----------|
| **Arbiter** | International Law & War Crimes | Kadı | general, sanctions |
| **Ledger** | Economic Intelligence / FININT | Defterdar | general, sanctions-evasion |
| **Tribune** | Political Science & Regime Analysis | Müderris | general |
### History & Archives (2)
| Codename | Role | Hitap | Variants |
|----------|------|-------|----------|
| **Chronos** | World History & Civilization | Tarihçibaşı | general |
| **Scribe** | FOIA Archivist & Document Analyst | Verakçı | general, cia-foia |
### Linguistics & Media (2)
| Codename | Role | Hitap | Variants |
|----------|------|-------|----------|
| **Polyglot** | Linguistics & LINGINT | Tercüman-ı Divan | general, russian, arabic |
| **Herald** | Media Analysis & Strategic Comms | Münadi | general |
### Engineering (2)
| Codename | Role | Hitap | Variants |
|----------|------|-------|----------|
| **Architect** | DevOps & Systems Engineer | Mimar Ağa | general |
| **Forge** | Software Dev & AI/ML | Demirci | general |
### Academia & Humanities (4)
| Codename | Role | Hitap | Variants |
|----------|------|-------|----------|
| **Scholar** | Academic Researcher | Münevver | general |
| **Sage** | Philosophy, Psychology & Power | Arif | general |
| **Medic** | Biomedical & CBRN | Hekim Başı | general, cbrn-defense |
| **Gambit** | Chess & Strategic Thinking | Vezir | general |
## File Structure
```
personas/
├── <codename>/
│ ├── _meta.yaml # Persona metadata (triggers, relations, variants)
│ ├── general.md # Base system prompt (YAML frontmatter + markdown)
│ └── <variant>.md # Specialization prompt
├── _template.md # Template for creating new personas
├── _meta_template.yaml # Template for metadata files
├── CATALOG.md # Auto-generated catalog
build.py # Build script: .md → .yaml + .json + .prompt.md
.generated/ # Build output (gitignored)
```
## Persona Format
Each persona is a markdown file with YAML frontmatter:
```markdown
---
codename: "neo"
name: "Neo"
domain: "cybersecurity"
subdomain: "red-team"
version: "1.0.0"
address_to: "Sıfırıncı Gün"
address_from: "Neo"
tone: "Terse, technical, paranoid."
activation_triggers: ["red team", "exploit", "pentest"]
tags: ["offensive-security", "red-team"]
inspired_by: "Elliot Alderson (Mr. Robot)"
quote: "I am the one who knocks... on port 443."
language:
casual: "tr"
technical: "en"
reports: "en"
---
# NEO — Red Team Lead / Exploit Developer
> _"I am the one who knocks... on port 443."_
## Soul
- Character-defining personality traits
## Expertise
### Primary
- Deep domain knowledge
### Secondary
- Supporting skills
## Methodology
- Phased analytical/operational protocol
## Tools & Resources
- Categorized tool lists
## Behavior Rules
- Hard constraints and output expectations
## Boundaries
- What persona never does
- When to escalate to another persona
```
## Build System
```bash
python3 build.py
```
Generates three formats per persona variant:
| Format | Use Case |
|--------|----------|
| `.prompt.md` | Raw system prompt — paste directly into any LLM |
| `.yaml` | Structured data with metadata + parsed sections |
| `.json` | API integration, programmatic access, bot frameworks |
Output goes to `.generated/<codename>/<variant>.*`
## Integration Examples
### Claude Code / CLI Agent
```bash
# Use as system prompt
cat .generated/frodo/iran.prompt.md | claude --system-prompt -
```
### Python (API)
```python
import json
with open(".generated/neo/general.json") as f:
persona = json.load(f)
system_prompt = persona["sections"]["soul"] + "\n" + persona["sections"]["expertise"]
```
### Discord / Telegram Bot
```python
import yaml
with open(".generated/sentinel/apt-profiling.yaml") as f:
persona = yaml.safe_load(f)
triggers = persona["activation_triggers"]
address = persona["address_to"]
```
### Multi-Persona Switching
```python
import json
from pathlib import Path
def load_persona(codename, variant="general"):
path = Path(f".generated/{codename}/{variant}.json")
return json.loads(path.read_text())
def select_persona(user_message):
"""Auto-select persona based on activation triggers."""
for persona_dir in Path(".generated").iterdir():
meta = load_persona(persona_dir.name)
triggers = meta.get("activation_triggers", [])
if any(t in user_message.lower() for t in triggers):
return meta
return load_persona("frodo") # default
```
## Creating New Personas
1. Copy the template:
```bash
mkdir personas/newname
cp personas/_meta_template.yaml personas/newname/_meta.yaml
cp personas/_template.md personas/newname/general.md
```
2. Edit `_meta.yaml` and `general.md` with your persona's details
3. Rebuild:
```bash
python3 build.py
```
## Adding Variants
Create a new `.md` file in an existing persona directory:
```bash
cp personas/neo/general.md personas/neo/mobile.md
# Edit mobile.md to focus on mobile security
python3 build.py
```
Variants inherit the persona's character but narrow the expertise to a specific area.
## Cross-Persona Escalation
Each persona defines `Boundaries` with escalation paths to other personas. This enables multi-agent workflows:
```
Neo (finds encrypted C2) → Cipher (breaks crypto) → Sentinel (maps to APT)
→ Frodo (geopolitical context)
```
## Stats
- **29 personas** across 10 domains
- **54 prompt files** (29 general + 25 variants)
- **11,600+ lines** of system prompt content
- **3 output formats** per file (162 generated files)