No short list proves a package safe. Use these dimensions to decide whether to reject it, seek qualified review, or test a pinned version under tighter permissions.
Reviewed 31 August 2026. A Claude Skill can include executable scripts, references, templates, and other resources—not only behavioral instructions. Review the whole filesystem package using Anthropic's current Agent Skills overview as the product reference.
The left column shows evidence that needs investigation or rejection. The right column shows stronger evidence—not a guarantee. Record what you reviewed, the exact version, required access, decision owner, and expiry date.
A delivery channel, star count, badge, or familiar name is not proof. Trace the package to an authoritative maintainer, pin an exact revision, and compare a hash or signature where one is available. Treat private attachments as untrusted input.
Executable files are supported skill-package contents, so their presence is not automatically malicious. Review source, inputs, outputs, subprocesses, dependencies, writable locations, error paths, and required tools. Reject download-and-execute patterns.
List the exact files, directories, tools, and data classes the skill needs. A general writing skill has no credible reason to read credentials. Do not place secrets or production records in test fixtures.
Reject instructions that hide material actions, suppress permission prompts, fabricate success, or weaken policy. Normal concise output is different from concealing a consequential tool action.
Claude Code may have network-capable tools subject to permissions; API skill containers have no network access. For any external connection, document why it is needed, allowlist the destination, minimize transmitted data, and pin or validate downloaded content.
Do not approve code you cannot assess. Obfuscation and unexpected generated binaries are severe signals, but complexity alone is not proof of malicious intent. Escalate unfamiliar languages or dependencies to a qualified reviewer.
Use model review as a second pass only: after manual inventory, ask Claude to identify file reads, execution, dependencies, tool calls, network paths, hidden instructions, and missing controls. Models can miss, misread, or invent findings. Never paste secrets or proprietary code into an unapproved service, and never let “looks safe” authorize installation.
Legitimate skills may use instructions, scripts, files, and tools. Safer packages have clear provenance, narrow purpose, inspectable contents, least privilege, isolated tests, explicit approval for consequential actions, and monitoring plus removal. None of those alone is a guarantee.