How rammaruboina-rgb/iceberg works
Iceberg combines a local bug-bounty assistant, a safe file-inventory helper, and a static inventory-review dashboard. The main workflow is submitting bounty context to the localhost agent, finding relevant local files, optionally asking an LLM for help, and saving reviewable results. The agent also exposes file, manifest, and conversation features. Source excerpts are partial, so links for unsampled endpoint behavior and persistence are omitted where their wiring is not established.
Main entry points: agent.py Local API · bounty_file_agent.py Inventory CLI · script.js Inventory dashboard · index.html Dashboard page
Components 13
Bounty assistant
- Local API
bounty-agent/agent.py - Bounty intake
bounty-agent/agent.py - Local file search
bounty-agent/agent.py - Assistant chat
bounty-agent/agent.py - Agent results
bounty-agent/results
Safe inventory
- Inventory CLI
agent/bounty_file_agent.py - File records
agent/bounty_file_agent.py - Inventory reports
agent/reports
Review dashboard
- Inventory dashboard
website/script.js - Inventory metadata
website/bounty-inventory.json - Dashboard page
website/index.html
Other components
- Local reviewer
- OmniRouter
Connections 14
- Local reviewer → Local API: submits requests
- Local API → Bounty intake: dispatches intake
- Bounty intake → Local file search: searches context
- Local API → Local file search: searches files
- Local API → Agent results: stores results
- Local API → Assistant chat: dispatches chat
- Assistant chat → Local file search: retrieves context
- Assistant chat → OmniRouter: sends prompt
- Inventory CLI → File records: collects metadata
- Inventory CLI → Inventory reports: writes reports
- Local reviewer → Inventory CLI: runs commands
- Local reviewer → Dashboard page: opens dashboard
- Dashboard page → Inventory dashboard: loads script
- Inventory dashboard → Inventory metadata: fetches metadata