Skillv1.2.0

ClawScan security

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

BenignMar 9, 2026, 11:05 PM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: Instruction-only skill that evaluates prospect lists you provide; its requested access and runtime instructions align with the stated scoring purpose.
Guidance: This skill is an offline scoring framework that operates on prospect lists you provide and does not itself access LinkedIn or external services. Before using: avoid pasting sensitive personal data (SSNs, private credentials, medical info), review any outreach text the agent generates before sending, and confirm compliance with LinkedIn and privacy rules if you use the results for outreach. If you plan to buy the paid pack, verify the vendor/site independently before sharing payment details.

Purpose & Capability: okName/description (scoring LinkedIn prospects) match the SKILL.md content. The skill is a decision framework that operates on prospect lists you supply and does not request unrelated credentials or tools.
Instruction Scope: okSKILL.md instructs the agent to read a user-provided prospect list (CSV/plain text) and apply explicit scoring rules. It does not instruct web scraping, reading system files, accessing environment variables, or transmitting data to third-party endpoints. It does note that sending messages would use the agent's normal messaging capabilities (user review required).
Install Mechanism: okNo install spec and no code files present (instruction-only). Nothing will be written to disk or downloaded as part of the skill itself.
Credentials: okThe skill declares no required env vars, credentials, or config paths. That matches its described purpose of locally scoring user-supplied prospect lists.
Persistence & Privilege: okSkill is not marked always:true and requests no elevated privileges or persistent system changes. It can be invoked autonomously per platform defaults, which is normal for skills.