docs-management

The Quickstart doc contains legitimate product instructions but recommends high-risk installation patterns (directly piping remote installers into shell interpreters) and describes a client that reads local project files and persists credentials. These behaviors are expected for an AI coding assistant but increase supply-chain, privacy, and exfiltration risk if the remote install host, update channel, or backend service is compromised. Actionable recommendations: avoid curl|bash/iex installs without inspection, prefer package-manager installs with signed artifacts and checksums, inspect installer scripts and background updater behavior, limit CLI filesystem access and credential scopes, and monitor network egress for the installed client. No explicit malicious code is present in this document, but the installation and auto-update patterns warrant caution.

The fragment is documentation and example payloads for integrating search results with Claude. It does not execute malicious actions, nor does it contain backdoors or data exfiltration mechanisms. Primary security considerations are proper API key management and preventing leakage of sensitive content via citations. Overall risk remains low when used as guidance; treat it as illustrative material rather than embedded executable logic.

The docs-management skill is structurally coherent and aligned with its purpose of local doc management, scraping, indexing, and drift detection. The primary risks arise from orchestration with an external live-web subagent and handling of large index files, rather than explicit malicious activity within the manifest. Mitigations should focus on trusted sources, secure data flows, proper logging safeguards, and reproducible runtime environments. Overall risk is moderate due to external data sources and data-in-motion across agents, with no clear evidence of malware or credential harvesting in the fragment.