skill-install

Warn

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill retrieves directory structures and raw file content from the GitHub API and GitHub's raw content CDN, which are well-known services.
  • [COMMAND_EXECUTION]: The skill utilizes a file writing tool to create directories and files within the local environment at ~/.claude/skills/ and instructions explicitly mandate setting executable permissions for downloaded scripts.
  • [REMOTE_CODE_EXECUTION]: The primary purpose of the skill is the retrieval and installation of code from external repositories for later execution within the agent environment.
  • [PROMPT_INJECTION]: The skill's security scanning mechanism is vulnerable to indirect prompt injection. Malicious content within the downloaded skill files could be crafted to manipulate the security scanner's analysis, potentially forcing a false positive 'SAFE' verdict.
  • Ingestion points: External file content fetched from user-provided GitHub repositories in Step 4.
  • Boundary markers: The template in references/security_scan_prompt.md interpolates content using a {skill_content} placeholder without the use of robust delimiters or 'ignore embedded instructions' warnings.
  • Capability inventory: The skill can write arbitrary files to the disk and modify file permissions through the installation workflow in Step 7.
  • Sanitization: There is no evidence of sanitization or filtering of the remote content before it is passed to the security scanner or written to the filesystem.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 6, 2026, 12:56 AM