Skillv0.1.2

ClawScan security

entrepreneur-skill · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

BenignApr 17, 2026, 4:19 PM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's files and instructions are consistent with a founder/copilot persona that generates weekly reviews and workflow guidance; it does not request secrets or install external code and the contained script is a simple local report generator.
Guidance: This skill appears internally consistent and low-risk: it contains a persona, a set of reference docs, and a small Python script that reads a JSON file and writes a markdown report. Before installing, confirm your agent runtime's network and tool policies — the persona declares optional network/onchain settings and allows tools like curl/WebFetch, so verify the agent won't send sensitive data to external gateways unless you explicitly enable the optional integrations. Also review any inputs you pass to the included script (JSON files) to avoid accidentally writing or exposing confidential data in generated reports.

Purpose & Capability: okName/description (founder partner) align with provided persona, references, and a small automation script. No unrelated environment vars, binaries, or installs are required. Optional external integrations are declared but not required.
Instruction Scope: okSKILL.md instructs local use of the included Python script to generate markdown from provided JSON and references internal docs to run workflows. There are no runtime instructions to read arbitrary host files, access unrelated credentials, or transmit data to unknown remote endpoints.
Install Mechanism: okNo install spec is present (instruction-only plus a small included script). Nothing is downloaded or extracted from external URLs; risk from installation is minimal.
Credentials: noteThe skill requests no environment variables or credentials. The persona.json includes optional network-related fields (social.gateway, onchain, a2a) and additionalAllowedTools (curl/WebFetch) which enable outbound calls in principle, but SKILL.md does not instruct using them and no secrets are declared. This is notable but not necessarily incoherent.
Persistence & Privilege: okalways is false and model invocation is permitted (normal). The skill does not request persistent privileges or modify other skills' configs; its only runtime persistence is writing generated reports to a specified output path.