Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed pastebin/URL-shortener integration that sends user-chosen non-secret content to a fixed hosted service.
Install only if you are comfortable sending selected handoff notes or URLs to the hosted 呵.tw service. Do not use it for secrets, private logs, credentials, regulated personal data, or content that should stay local.
Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.
Never send:
- API keys
- access tokens
- passwords
- private cookies
- SSH keys
The skill instructs the agent to use a bundled CLI to communicate with a hosted external service, which is a network-capable action, but the manifest declares no explicit tool scope or permission boundary. That creates an authorization and review gap: an agent may invoke outbound network behavior without a machine-readable restriction, increasing the chance of unintended data egress or unsafe automatic use.
The create-paste command reads arbitrary content from --content or stdin and posts it to a third-party remote service, but the code contains no warning, confirmation, secrecy check, or policy guard. In an agent skill whose purpose is to offload context and handoff notes, this materially increases the chance that sensitive prompts, credentials, internal data, or user secrets are exfiltrated to an external service by mistake.
Dynamic getattr() with a non-literal attribute name can access arbitrary object attributes, potentially bypassing access controls.
raise SystemExit("content required (use --content or --stdin)")
payload = {"content": content}
for key in ["title", "format", "summary", "parent_slug"]:
value = getattr(args, key)
if value:
payload[key] = value
if args.tags:
The argument parser description uses Chinese text ('Use 呵.tw for token-saving agent handoff') with a non-English service name and does not provide any language selection or alternative locale. This can constitute a natural-language policy issue when a skill presents a fixed language/locale choice without opt-in or documented justification.