plugin-write

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The query-registry.mjs script performs network requests to fetch a plugin registry index from the vendor's repository on GitHub (oh-my-dsh/dsh-plugin-registry). This is a functional requirement for checking naming conflicts against reviewed community registrations. The script uses a well-known service and targets resources belonging to the skill author.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data in the form of naming manifests (dsh-plugin.naming.json) provided by users or external plugin projects.
  • Ingestion points: The validate-names.mjs and query-registry.mjs scripts read local manifest files from the workspace.
  • Boundary markers: The scripts treat the input as structured JSON rather than natural language instructions.
  • Capability inventory: The skill's scripts utilize readFile for local manifests and fetch for the remote registry; test scripts use spawn to verify CLI behavior.
  • Sanitization: The skill implements robust sanitization by validating all input against a strict naming policy and JSON schema (plugin-naming.schema.json) before use.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 05:59 PM
Security Audit — agent-trust-hub — plugin-write