Skillv1.1.1

ClawScan security

Massive Financial Connector · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

SuspiciousMar 5, 2026, 7:12 PM

Verdict: suspicious
Confidence: medium
Model: gpt-5-mini
Summary: The skill mostly does what its description says (query Massive/Polygon), but the package metadata omits required environment/binaries and the start script relies on a local 'uvx' runner that will fetch and run code from GitHub at runtime — these mismatches and the remote-execution path are concerning and worth manual review before installing.
Guidance: This skill appears to implement a legitimate Massive/Polygon connector, but there are three red flags you should address before installing or running it: (1) The registry metadata does not list MASSIVE_API_KEY or required binaries (curl, python3, uvx) — verify and supply only the minimal credentials needed. (2) start-mcp-server.sh calls a local uvx binary with a git+ URL which will fetch and execute code from GitHub at runtime — inspect what uvx does and manually review the GitHub repo (massive-com/mcp_massive@v0.4.0) before allowing it to run. (3) The scripts source your ~/.zshrc which can execute arbitrary shell code and expose other env variables; consider removing that line or running the scripts in a controlled environment. Recommended steps: run the scripts manually in an isolated VM/container, confirm the uvx binary's origin and behavior (or install a vetted alternative), verify the remote repo contents, and ensure your MASSIVE_API_KEY is stored and used securely (not committed or uploaded). If you cannot audit uvx and the GitHub repo, do not run the start script with network access.

Review Dimensions

Purpose & Capability: concernThe skill's stated purpose (Massive/Polygon market-data connector) matches the scripts which call api.massive.com endpoints. However the registry metadata declares no required env vars or binaries while SKILL.md and scripts require MASSIVE_API_KEY and expect curl, python3, and a local uvx binary. The missing declarations are incoherent with the claimed functionality.
Instruction Scope: concernSKILL.md instructs the agent to run the provided scripts and to start the official MCP server. The scripts source the user's ~/.zshrc (silently), read MASSIVE_API_KEY, call api.massive.com via curl, and the server script execs a local uvx binary that will fetch/run code from a GitHub repo. Sourcing ~/.zshrc can execute user dotfile content and may expose or run unexpected state; the uvx-based remote fetch potentially downloads and executes code beyond the local files.
Install Mechanism: concernThere is no install spec, but start-mcp-server.sh relies on an external runner ($HOME/.local/bin/uvx) invoked with a git+https://github.com/... URL which will pull code from GitHub at runtime. This is effectively a remote download-and-execute step that is not declared or constrained by an install block; whether it is safe depends entirely on the uvx tool and the remote repo's integrity.
Credentials: concernRequesting MASSIVE_API_KEY is appropriate for a Massive/Polygon connector, but the skill metadata omitted that requirement. The scripts also implicitly rely on curl and python3. Additionally, the scripts source ~/.zshrc which may expose other environment variables or execute arbitrary shell code — this is not justified by the stated purpose and increases risk of unintended side-effects or secret access.
Persistence & Privilege: okThe skill does not request always:true and does not modify global agent configuration in the provided files. There is no install spec that writes persistent system-wide artifacts in the package itself. The main privilege/risk comes from the runtime behavior (uvx fetching remote code), not from declared persistent privileges.