How assisgb/pondersec works
PonderSEC is a Django platform for cybersecurity-focused LLM evaluation. Its main workflows generate model answers to questions, then collect human or JudgeAI evaluations and present comparisons. The README documents public chat with cross-evaluation, researcher-managed questions and models, external evaluation forms, and a separate admin panel. The tree and excerpts support the web entry points and data model; the README names internal LLM and JudgeAI modules that are absent from the supplied tree, so those components use the enclosing app path and their wiring is shown only where documented.
Main entry points: responsegenerator/views.py Product workflows · models.py Application database · urls.py Project URL routing · usuarios/views.py Researcher accounts
Components 22
Web entry and accounts
- Project URL routing
pondersec/urls.py - Researcher accounts
usuarios/views.py - Account routes
usuarios/urls.py - Product routes
responsegenerator/urls.py - Product workflows
responsegenerator/views.py
Research workflows
- Question bank
responsegenerator/views.py - Private model setup
responsegenerator/views.py - Answer generation
responsegenerator
Evaluation
- Manual review
responsegenerator/views.py - JudgeAI evaluation
responsegenerator - Specialist forms
responsegenerator/views.py - Scores and comparison
responsegenerator/views.py
Public chat and admin
- Public chat
responsegenerator/views.py - Public cross-evaluation
responsegenerator - Admin panel
responsegenerator/views.py
Persistent data
- Application database
responsegenerator/models.py
Other components
- Visitor
- Researcher
- External specialist
- Panel admin
- LLM providers
- Email service
Connections 36
- Visitor → Project URL routing: opens chat
- Researcher → Project URL routing: opens app
- External specialist → Project URL routing: opens form
- Panel admin → Project URL routing: opens panel
- Project URL routing → Account routes: routes account
- Project URL routing → Product routes: routes product
- Account routes → Researcher accounts: dispatches
- Product routes → Product workflows: dispatches
- Researcher accounts → Application database: reads/writes accounts
- Researcher accounts → Email service: sends verification
- Product workflows → Question bank: handles questions
- Product workflows → Private model setup: configures models
- Product workflows → Answer generation: starts generation
- Product workflows → Manual review: handles review
- Product workflows → JudgeAI evaluation: runs judging
- Product workflows → Specialist forms: manages forms
- Product workflows → Scores and comparison: serves scores
- Product workflows → Public chat: serves chat
- Product workflows → Public cross-evaluation: starts evaluation
- Product workflows → Admin panel: serves admin
- Question bank → Application database: reads/writes questions
- Private model setup → Application database: saves model config
- Answer generation → LLM providers: requests answers
- Answer generation → Application database: stores answers
- Manual review → Application database: stores ratings
- JudgeAI evaluation → LLM providers: requests judgments
- JudgeAI evaluation → Application database: stores scores
- Specialist forms → Application database: reads/writes evaluations
- External specialist → Specialist forms: submits ratings
- Scores and comparison → Application database: reads scores
- Public chat → Application database: reads public models
- Public chat → LLM providers: requests responses
- Public chat → Application database: stores chat
- Public cross-evaluation → LLM providers: requests judgments
- Public cross-evaluation → Application database: reads metrics; stores ratings
- Admin panel → Application database: manages public config