Finding metric we got from the benchmark
we had a major upgrade on @bountyforge_ hunting sequence had massive improvements on the bench99 test from 13 findings from past version to 70/99 confirmed on the new version with the red cyber 1 model ,
112
Bountyforge ai retweeted
$1,000,000 high. $50,000 medium. $10,000 low.
Safe Foundation pays those tiers in USDC on Ethereum mainnet for smart contract bugs: theft, permanent lockups, and anything that forces an urgent redeploy.
Submit to [email protected].
docs.safefoundation.org/secu…
👀
👇👇👇
Cairn: github.com/oritera/Cairn
CyberStrike: github.com/CyberStrikeus/cyb…
PentAGI: github.com/vxcontrol/pentagi
CyberStrikeAI: github.com/AIPentest/CyberSt…
PentestGPT: github.com/GreyDGL/PentestGP…
PentestAgent: github.com/GH05TCREW/pentest…
AutoPentester: github.com/YasodGinige/AutoP…
VulnBot: github.com/KHenryAegis/VulnB…
RedAmon: github.com/samugit83/redamon
MAPTA: github.com/arthurgervais/map…
Strix: github.com/usestrix/strix
Shannon: github.com/KeygraphHQ/shanno…
BugTraceAI: github.com/BugTraceAI/BugTra…
Nebula: github.com/BerylliumSec/nebu…
hackingBuddyGPT: github.com/IPA-Lab/HackingBu…
HexStrike AI: github.com/0x4m4/hexstrike-a…
CAI, now archived: github.com/aliasrobotics/CAI
2
73
New pwn skill by our team lead
@gabson0x
github.com/BountyForge-xyz/p…
Bountyforge ai retweeted
The @plether_fi audit competition launches in less than 3 days.
💰 $80,000 reward pool
📅 Runs Nov 2 – Nov 15.
Open to all registered researchers who meet the eligibility rules.
Program details, and when it goes live: immunefi.com/audit-competiti…
The @plether_fi audit competition launches in less than 3 days.
💰 $80,000 reward pool
📅 Runs Nov 2 – Nov 15.
Open to all registered researchers who meet the eligibility rules.
Program details, and when it goes live: immunefi.com/audit-competiti…
We’re kicking off BENCH99 by BountyForge ai
Testing different AI models on their ability to find real vulnerabilities, measuring confirmed findings, discovery rates, and cost.
99 bugs. Multiple models. One benchmark.
Let’s see which models actually deliver.
#BENCH99
We are the ai pentest engineer for hunting , nothing gets pass our sight
Replying to @bountyforge_
@bountyforge_ Found a Fastly edge WAF bypass on target.com using sqli & x-forwarded headers
'1 OR 1=1' returns 406, but the equivalent '1 OR 1' returns 200.
Same endpoint. Different filtering outcome.
A reminder that signature-based WAF rules can be bypassed when they match payload syntax instead of malicious behavior.
1
87
Akamai Waf bypass tip found by @bountyforge_ ai
Normal curl /browser on the expample.com =403
But using AWS origin ip returned 301
🤖 Made with AI
Need POC for CVE’s?
Currently 7 cve poc attached and tested e2e
github.com/BountyForge-xyz/C…
The death of manual vulnerability hunting ?
We set up an article of how the new wave of SI works in pentesting
bountyforge.xyz/blog/the-dea…
5 major types of RCE patterns
^Unauthenticated upload exec
^authenticated deserialization
^command injection
^ expression language injection
1
120
What's in the box today:
• Explorer source fetched, reconciled against deployed bytecode before it's trusted
• Slither + protocol model + proxy evidence, confidence-gated at ≥2 independent signals
• Explicit fork status on every run never a silent static-only
we are working on integrating web3 scans
Slither + source-vs-bytecode reconcile + a confidence gate that requires ≥2 independent signals before a surface gets promoted. Source is untrusted until it MATCHes deployed bytecode.
Fork replay is gated, not faked. Every run stamps an explicit status and an RPC-blind witness hash. If the fork can't run, the report says so. No zero-day theater without a witness.
we are working on integrating web3 scans
Slither + source-vs-bytecode reconcile + a confidence gate that requires ≥2 independent signals before a surface gets promoted. Source is untrusted until it MATCHes deployed bytecode.
Fork replay is gated, not faked. Every run stamps an explicit status and an RPC-blind witness hash. If the fork can't run, the report says so. No zero-day theater without a witness.
5/6 — bountyforge
github.com/Gabson0x/bountyfo…
~400 ⭐️
Built around parallel Claude agents for smart-contract auditing:
⁃ EVM
⁃ Move
⁃ Solana
⁃ TRON
⁃ Plus web/API security.
If you're hunting Web3 bounties, this one is worth exploring.



















