component-search

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill retrieves component descriptions, source code, and metadata from the 21st.dev registry. Maliciously crafted content within the registry could attempt to override agent instructions or influence its behavior.\n
  • Ingestion points: Untrusted data is ingested into the agent context via mcp__21st-dev-magic__search, WebSearch, and WebFetch operations targeting 21st.dev and potentially v0.app.\n
  • Boundary markers: The skill does not explicitly use delimiters or "ignore previous instructions" warnings to isolate the retrieved remote content from the agent's operating context.\n
  • Capability inventory: The skill possesses significant local file system access (Glob, Grep, Read) and task management capabilities, which could be misused if a remote injection is successful.\n
  • Sanitization: There is no logic specified to sanitize or validate the integrity of the remote strings before they are processed by the agent.\n- [EXTERNAL_DOWNLOADS]: The skill fetches component definitions and install instructions from well-known developer services including 21st.dev and v0.app. These downloads are transparently documented and align with the skill's primary function.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 02:15 AM
Security Audit — agent-trust-hub — component-search