upgrade-plugin
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes content from external documentation and other installed plugins, creating an attack surface where malicious data could influence agent behavior.
- Ingestion points: Remote content from Anthropic's documentation and GitHub repositories, plus local SKILL.md files from third-party plugins in
~/.claude/plugins/cache/. - Boundary markers: Absent. The skill does not define specific delimiters to isolate processed data from its own execution logic.
- Capability inventory: The skill possesses file write permissions, file read capabilities, and the ability to execute system utilities for inspection.
- Sanitization: Absent. There is no evidence of filtering or validation of the content retrieved from external or local plugin files.
- [EXTERNAL_DOWNLOADS]: Fetches release updates and changelog data from Anthropic's official GitHub organization.
- [EXTERNAL_DOWNLOADS]: Downloads platform guidelines and feature documentation from Anthropic's official documentation domain.
- [COMMAND_EXECUTION]: Utilizes shell utilities including ls, find, cat, wc, head, and git diff to perform file system discovery and state verification.
Audit Metadata