akita-xp-rules
Pass
Audited by Gen Agent Trust Hub on Jun 20, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local shell script (
scripts/version-check.sh) at the start of each session to verify the current version of the toolset. - [EXTERNAL_DOWNLOADS]: The version-check script performs network requests to the official npm registry (
registry.npmjs.org) to fetch the latest version information for thexp-stackpackage. This is a standard update mechanism targeting a well-known service. - [PROMPT_INJECTION]: The skill includes instructions to suppress AI attribution in git commit messages (e.g., omitting 'Co-Authored-By: Claude'). While framed as a stylistic preference for clean git history, this hides the agent's contribution to the codebase.
- [PROMPT_INJECTION]: Category 8: Indirect Prompt Injection surface.
- Ingestion points: The skill designates
CLAUDE.mdat the project root as the 'absolute source of truth'. - Boundary markers: None. The agent is instructed to read the file entirely and follow it 'as if they were your own' instructions.
- Capability inventory: The agent is authorized to perform git commits, run shell scripts, and manage project architecture.
- Sanitization: There are no instructions to sanitize or validate the content of the
CLAUDE.mdfile before adopting its instructions.
Audit Metadata