improve-skill
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to clone the vendor's repository (maragudk/fabrik) to a temporary directory if a local version is not detected.
- [COMMAND_EXECUTION]: The instructions involve executing Git commands to manage code changes, specifically using
git worktree add,git clone,git push, andgit committo manage improvements. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted conversation history to generate summaries and code modification suggestions, creating a potential surface for indirect prompt injection.
- Ingestion points: Current conversation history is parsed to identify 'signals' (SKILL.md, Step 1).
- Boundary markers: No explicit delimiters are used to wrap the conversation history during processing.
- Capability inventory: The agent has the ability to execute Git commands, modify files within the repository scope, and perform network operations to push branches and create PRs.
- Sanitization: The skill implements a mandatory human review step (SKILL.md, Step 2) where all findings and proposed changes must be presented to and accepted by the user before the assistant prepares the working copy or ships changes.
Audit Metadata