Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Polygon Agents SDK
v1.0.0Complete Polygon agent toolkit. Session-based smart contract wallets (Sequence), token ops (send/swap/bridge/deposit via Trails), ERC-8004 on-chain identity...
⭐ 1· 347·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Name, description, and the CLI commands in SKILL.md consistently describe a Polygon agent toolkit (wallet/session creation, token ops, ERC‑8004 registration). The declared environment variables in the docs (SEQUENCE_PROJECT_ACCESS_KEY, SEQUENCE_INDEXER_ACCESS_KEY, optional Trails keys) are appropriate for the described functionality.
Instruction Scope
The instructions require running a third‑party CLI (npm install -g github:0xPolygon/polygon-agent-kit) that will: create and store private keys, auto-open a Cloudflare Quick Tunnel, save session blobs to /tmp, and insist you transmit full approval URLs to users. Those actions are expected for a session-based wallet flow, but they broaden the agent's runtime surface (network-exposed tunnels, temporary plaintext blobs, local storage of private keys) and could enable attack vectors if the underlying code or downloaded binaries are malicious or tampered with.
Install Mechanism
SKILL.md instructs installation from a GitHub repo via npm (github:0xPolygon/polygon-agent-kit) — a common pattern but still fetches code at install time. More concerning: the CLI auto-downloads a cloudflared binary into ~/.polygon-agent/bin/ on first use. Auto-downloading and executing binaries from the network increases risk unless the source and integrity checks (signatures/checksums) are explicit. The registry metadata itself had no install spec or homepage listed, which reduces traceability.
Credentials
Required environment variables listed in SKILL.md (SEQUENCE_PROJECT_ACCESS_KEY and SEQUENCE_INDEXER_ACCESS_KEY) are directly relevant to wallet/session operations. Optional vars (TRAILS_API_KEY, TRAILS_TOKEN_MAP_JSON, debug flags) also make sense. The skill will output and ask you to save private keys and stores encrypted material in ~/.polygon-agent/ — expected, but these are high‑sensitivity secrets and the workflow involves temporary plaintext blobs in /tmp and URLs that must be copied exactly.
Persistence & Privilege
The skill is instruction-only and does not request always:true or other elevated platform privileges. However, the CLI behavior described will create persistent files under ~/.polygon-agent/ (encrypted storage and a bin folder for cloudflared) and write temporary files under /tmp; it also spawns a network tunnel. Those are normal for a CLI wallet but constitute persistent disk and network state that you should be willing to host on the machine used.
Scan Findings in Context
[no-regex-findings] expected: The scanner found no code files to analyze (the skill is instruction-only). This is expected given the package is not embedded in the registry entry — it directs users to install a GitHub-hosted npm package instead. Absence of findings does not mean the upstream package or auto-downloaded binaries are safe.
What to consider before installing
This skill appears to implement what it claims, but it performs sensitive operations (generates/saves private keys, auto-downloads/executes cloudflared, exposes a public tunnel and temporary blobs). Before installing or running it: 1) Verify the upstream repo (https://github.com/0xPolygon/polygon-agent-kit) is authentic and review its code — especially any cloudflared download logic and where binaries are fetched from. 2) Prefer installing and running in an isolated environment (VM/container) rather than a primary machine. 3) Do not paste or share private keys or session blobs publicly; treat SEQUENCE_* keys as secrets. 4) If you must use it on a workstation, verify any downloaded binary checksums/signatures and consider using your own Cloudflare/connector setup instead of auto-tunnel. 5) If you lack comfort auditing the code, ask for a signed release or a package from a verifiable publisher before proceeding.Like a lobster shell, security has layers — review code before you run it.
latestvk977dgn0382ypw05txs9rfrsyn81xqve
License
MIT-0
Free to use, modify, and redistribute. No attribution required.