add-community-extension

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) due to its core workflow of processing external, user-provided content.
  • Ingestion points: The agent reads content from untrusted GitHub issue submissions (issue URL/number) and fetches metadata from third-party repositories (extension.yml, README.md, and release tags).
  • Boundary markers: Absent. The instructions do not provide delimiters or guidance to ignore instructions that might be embedded within the extension's name, description, or repository content.
  • Capability inventory: The skill allows the agent to modify local files (catalog.community.json, extensions.md), execute shell commands (git, python3), push code to remote branches, and open pull requests to the upstream repository.
  • Sanitization: While the skill enforces structural validation for certain fields (regex for extension IDs and semver for versions), it lacks sanitization for free-text fields like the extension description, which are directly interpolated into the catalog and documentation.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 02:53 PM