update-rules
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes and displays instruction-based content from markdown files and templates, which could potentially host malicious prompts intended to influence agent behavior during analysis.
- Ingestion points: The skill reads .rules/*.md, AGENTS.md, CLAUDE.md, and template files from the plugin's template directory as documented in SKILL.md.
- Boundary markers: The process includes mandatory human-in-the-loop checkpoints, requiring users to review unified diffs and explicitly confirm changes before they are applied (SKILL.md).
- Capability inventory: The skill uses the Edit tool for surgical file modifications, the Write tool for new file creation, and executes the bin/project-diff-rules script and standard diff utility (SKILL.md).
- Sanitization: There is no automated sanitization of the content; the skill relies on the user's manual review of the proposed changes as the primary defensive layer.
- [COMMAND_EXECUTION]: The skill invokes a local script and system utilities to perform configuration audits.
- Evidence: SKILL.md specifies the execution of a bundled helper script (bin/project-diff-rules) by absolute path from the plugin installation directory, as well as the standard diff utility.
Audit Metadata