🔍 The Problem Statement
In this post, I disclose a logic vulnerability found in a Starknet staking protocol (specifically Attestation.cairo). The issue allows an attacker to extract STRK rewards without providing the mandatory BTC collateral.
🛠 The Technical Deep Dive: get_block_hash_syscall
The vulnerability stems from the improper handling of block hashes for off-chain verification.
Rust
// Simplified vulnerable snippet
let block_hash = get_block_hash_syscall(target_block_number).unwrap();
assert(provided_proof.anchor == block_hash, 'Invalid proof anchor');
The flaw is that the contract relies on a syscall to verify external data without a cryptographic link that prevents replaying or spoofing the block reference in a specific context. By manipulating the target_block_number or providing a stale but valid hash, an attacker can satisfy the assert condition.
🛡 Why it matters
Reward Extraction: Attackers can claim incentives meant for real stakers.
Economic Imbalance: Dilutes the value of the protocol for honest participants.
📉 The Human Factor: 48 Hours of Silence
I initially reached out to the team privately. The response was a dismissal, labeling the report as "AI slop". Despite providing a functional PoC on snforge and giving a 48-hour window for a technical review, I received no further response.
Note: Security is about the code, not the writing style. Dismissing a critical vulnerability based on a hunch puts the entire ecosystem at risk.
🚀 Proof of Concept
I've released a public repository with a full test suite demonstrating the exploit:
👉 https://github.com/rdin777/starknet-staking_audit/tree/main
🔚 Conclusion
Public disclosure was necessary to warn the community. If you are developing on Starknet, ensure your attestation logic is tightly coupled with cryptographic proofs that cannot be decoupled from their intended block state.
