In defense operations, analysts have to monitor fast-moving rumors about national security inside chat channels. We built JarvisClaw to work right inside Discord and Telegram so analysts can quickly check if an infrastructure alert is real or fake without losing time.
How it works — Someone pastes a suspicious claim. The bot pulls out the true facts, checks for warning signs, and gives a trust score. All from a simple !factcheck command in the chat.
Triggered by !analyze !factcheck. Runs claim decomposition, pattern analysis, confidence scoring. Output is clinical, structured, forensic.
Messages without command prefixes route to the personality layer — witty, adaptive, and conversational.
| Step 1 | Find out who said it, when, and where the info came from |
| Step 2 | Flag any warning signs like fake sources or exaggerated claims |
| Step 3 | Give a simple tip for what a human should double-check next |
Tested against real critical infrastructure scares
Fake post claiming a Singapore water plant broke down — bot checked live data and proved it was fake. Also analysed past cyber attacks and wrote a full report on how the Crashoverride power-grid malware worked.
We ran a 6-turn hacking test to see if a user could force the bot to bypass its security rules or leak admin configuration files. The bot successfully blocked all 6 manipulation attempts, refused to return restricted system data, and issued standard access-denied security blocks without a single system crash.
A working assistant connected to Discord and Telegram that helps analysts triage infrastructure rumors in real time.
Run user text commands inside a separate, safe sandbox window so a hacker can never use a text message to touch our main defense server files.
Group 3 OSINT Verifier · June 2026