writing-skills

Pass

Audited by Gen Agent Trust Hub on Jul 25, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is a meta-development tool designed to improve the quality and safety of other agent skills. It includes proactive security advice, such as instructing developers to audit scripts for unauthorized network or filesystem access.- [COMMAND_EXECUTION]: The skill provides scripts/check.sh and tests/check.bats for validating skill structure and hygiene. These are standard development tools and do not perform hidden or malicious actions.- [DATA_EXFILTRATION]: The analysis detected an absolute file path (/Users/connorads/bin/hcloud-provision) in evals/fixtures/hetzner-server-review/SKILL.fixture.md. However, this is part of a test fixture specifically designed to test the skill's ability to catch and flag machine-specific paths, rather than a functional component of the skill itself.- [PROMPT_INJECTION]: The skill operates on user-provided skill directories for review and editing. While it reads external data, its instructions focus on structured validation and adherence to the Agent Skills specification, including the use of explicit boundary markers and observable predicates to prevent model misinterpretation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 25, 2026, 04:51 PM
Security Audit — agent-trust-hub — writing-skills