writing-audit

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes a shell script evals/run-static-checks.sh which executes local commands and a Python script for testing purposes. These operations are intended for development environments and are not part of the runtime interaction with users.
  • [PROMPT_INJECTION]: The skill processes user-supplied text for auditing, creating a surface for indirect prompt injection. Ingestion point: The 'existing piece of writing' provided by the user (Step 5 of SKILL.md). Boundary markers: No explicit isolation or 'ignore embedded instructions' markers are defined. Capability inventory: The skill uses sub-agent dispatch and file-writing tools (SKILL.md). Sanitization: No content filtering is mentioned. This is a common property of text-auditing skills and is not indicative of malicious intent.
  • [NO_CODE]: The skill references several external resources that are missing from the provided package, such as the playbooks in references/playbooks/ and the sub-agent logic in references/subagent-dispatch.md. While this limits the review of the full implementation, the available files show no security risks.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 09:38 PM
Security Audit — agent-trust-hub — writing-audit