solana-auditor

Fail

Audited by Socket on Mar 23, 2026

7 alerts found:

Securityx3Obfuscated Filex2Malwarex2
SecurityMEDIUM
references/hacking-agents/invariant-agent.md

The fragment is not executable but constitutes a high-risk attacker-oriented blueprint. It should be treated as a potential supply-chain risk if included in a library or docs, requiring strict controls, access restrictions, and explicit guidance to avoid embedding exploit-oriented patterns in production code or educational materials without proper safeguards.

Confidence: 70%Severity: 70%
SecurityMEDIUM
references/hacking-agents/shared-rules.md

The provided fragment is a governance/guidance document with attacker-oriented framing. It does not perform malware or business logic activity, but its mindset could enable aggressive, non-evidence-based auditing workflows if adopted uncritically. Treat as a high governance risk and ensure defender-focused tooling and reviews to prevent weaponization of findings.

Confidence: 59%Severity: 75%
SecurityMEDIUM
SKILL.md

The skill is coherent with its stated purpose and shows no clear credential theft or covert exfiltration. However, it is a high-risk AI security-auditing capability: it orchestrates multi-agent smart-contract analysis using attack-vector guidance, processes untrusted code at scale, and retains shell/file-write abilities. The remote GitHub VERSION check is low-to-moderate supply-chain risk, but the main concern is the intentional enablement of AI-driven security review/exploit analysis.

Confidence: 89%Severity: 74%
Obfuscated FileHIGH
references/hacking-agents/economic-security-agent.md

This artifact is an explicit, high-risk attacker playbook for economic/DeFi exploitation. It enumerates realistic, actionable techniques (oracle manipulation, token behavior abuse, CPI substitution, lamport/dust manipulation, capacity exhaustion, atomic extraction) that can lead to direct financial loss or denial-of-service when target programs lack standard defenses. Projects using external oracles, CPIs, token features, or global caps should treat this as a credible threat model and implement mitigations: validate CPI program IDs, enforce oracle freshness/confidence/decimals, reload account state after CPI, account for actual received token amounts, favor per-user caps or rate limits, and avoid business logic that depends on mutable lamport values. The document itself is not executable malware but is highly actionable malicious guidance; remediation and threat modeling are recommended.

Confidence: 98%
Obfuscated FileHIGH
references/hacking-agents/math-precision-agent.md

The fragment is a high-risk attacker playbook for exploiting arithmetic and precision bugs in financial smart contracts (Rust/Solana). It contains actionable steps that would enable an attacker to find and weaponize rounding, truncation, cast, and overflow issues. Although not executable code, its presence in a repository is dangerous and warrants immediate review and removal or restriction; code implementing the described calculations should be audited for the enumerated defects (wrong rounding directions, unsafe 'as' casts, saturating ops misuse, disabled overflow checks, float use, and decimal mismatches).

Confidence: 98%
MalwareHIGH
references/hacking-agents/execution-trace-agent.md

This document is a high-risk attacker playbook describing multiple concrete execution-flow exploit techniques against Solana programs. It should be treated as malicious guidance: it identifies common programming mistakes (serialization issues, sentinel reliance, ignored CPI results, stale AccountInfo use, remaining_accounts misuse, account close/revive) and explains how to abuse them to steal funds or corrupt state. There is no benign library code here; inclusion of this content in a package (especially if coupled with test harnesses, example programs, or scripts) is a serious red flag. Remediation: audit programs for the exact patterns described, add strict validation, re-deserialize after CPI, validate remaining_accounts owners/layout, propagate CPI errors, and avoid trusting sentinel values.

Confidence: 45%Severity: 90%
MalwareHIGH
references/hacking-agents/periphery-agent.md

The submitted text is a deliberate attacker guide aimed at compromising supply-chain trust via subtle bugs in utility modules. It is malicious in intent and maps to realistic, high-impact attack patterns (PDA collisions, serialization corruption, CPI redirection, compute exhaustion). No concrete code is provided to prove exploitation, but the presence of this guidance in a project is a critical security concern that justifies immediate repository-wide audit and incident response.

Confidence: 75%Severity: 95%
Audit Metadata
Analyzed At
Mar 23, 2026, 09:25 PM
Package URL
pkg:socket/skills-sh/sanbir%2Fsolana-auditor-skills%2Fsolana-auditor%2F@ea17db6ec848c940256cc9b414bae9c894765ff2
Security Audit — socket — solana-auditor